From ell2 at live.se Tue Mar 1 20:05:13 2011 From: ell2 at live.se (Leif Linderstam) Date: Tue, 1 Mar 2011 21:05:13 +0100 Subject: [Box Backup] Workaround for struct packing problem on ARM processors (was: Another wish for 2011) In-Reply-To: References: Message-ID: Hi Chris, > go. If you choose this you might think about adding the write and read > functions to IOStream instead. I was thinking about doing that, but the > only way of doing it was to immediately call the virtual Write just as > Archive.h does it, and that was contrary to the design goal I had of > mimicking the properties of the packed struct implementation as far as > possible. (Oh, I really have a bad habit of writing long, complicated sentences. And probably bad English as well.) Thinking a little bit more about this it can be done efficiently in IOStream as well. It needs a bit of work in IOStream as well as in all its subclasses. If you want it that way I can make a proposed change for it, but with the usual disclaimer; it will take a while. Cheers, Leif From achim+box at qustodium.net Thu Mar 10 12:17:57 2011 From: achim+box at qustodium.net (Achim J. Latz) Date: Thu, 10 Mar 2011 13:17:57 +0100 Subject: [Box Backup] daily builds of trunk for ubuntu In-Reply-To: <87aahlkq9g.fsf@faui44a.informatik.uni-erlangen.de> References: <8762td12it.fsf@faui44a.informatik.uni-erlangen.de> <4D6677F2.2050408@qustodium.net> <87aahlkq9g.fsf@faui44a.informatik.uni-erlangen.de> Message-ID: <4D78C175.6000407@qustodium.net> Hello Reinhard: On 24/02/2011 23:43, Reinhard Tartler wrote: > I also wondered about this but unfortunatly, I came to the conclusion > that this won't be possible, at lesat not easily. The whole thing works > because I'm mirroring the SVN to bzr and build from there. The build > recipe has only access to the bzr revision number, but not (no longer) > to the svn revision. > > What I could do however is to encode the build date to the version > string. would this help? Not sure, because then users still have to check what SVN revision was "current" on a given date. But since there is apparently a way to run certain shell code ("to encode the date"), would it be possible to run a "svn info" (or whatever is needed) on the box backup repo and use the ouput of that command in encoding the version string for bzr? Thanks again for setting all this up, Achim -- Achim J. Latz, Qustodium Internet Security achim.latz at qustodium.net ? http://www.qustodium.net Data Encryption ? Backup Automatisation ? E-Mail Protection From siretart at tauware.de Thu Mar 10 14:25:43 2011 From: siretart at tauware.de (Reinhard Tartler) Date: Thu, 10 Mar 2011 15:25:43 +0100 Subject: [Box Backup] daily builds of trunk for ubuntu In-Reply-To: <4D78C175.6000407@qustodium.net> (Achim J. Latz's message of "Thu, 10 Mar 2011 13:17:57 +0100") References: <8762td12it.fsf@faui44a.informatik.uni-erlangen.de> <4D6677F2.2050408@qustodium.net> <87aahlkq9g.fsf@faui44a.informatik.uni-erlangen.de> <4D78C175.6000407@qustodium.net> Message-ID: <87vczruk60.fsf@faui44a.informatik.uni-erlangen.de> On Thu, Mar 10, 2011 at 13:17:57 (CET), Achim J. Latz wrote: > Hello Reinhard: > > On 24/02/2011 23:43, Reinhard Tartler wrote: >> I also wondered about this but unfortunatly, I came to the conclusion >> that this won't be possible, at lesat not easily. The whole thing works >> because I'm mirroring the SVN to bzr and build from there. The build >> recipe has only access to the bzr revision number, but not (no longer) >> to the svn revision. >> >> What I could do however is to encode the build date to the version >> string. would this help? > > Not sure, because then users still have to check what SVN revision was > "current" on a given date. But since there is apparently a way to run > certain shell code ("to encode the date"), would it be possible to run a > "svn info" (or whatever is needed) on the box backup repo and use the > ouput of that command in encoding the version string for bzr? Unfortunately, shell code is not possible. The available variables are ${time}, and ${revno}. The documentation is here: https://help.launchpad.net/Packaging/SourceBuilds/BzrBuilder The current recipe[1] reads like this: ,---- | # bzr-builder format 0.3 deb-version 0.12~~+trunk+r{revno} | lp:boxbackup | nest packaging lp:~siretart/+junk/boxbackup.packaging debian `---- This means the current version scheme is '0.12~~+trunk+r{revno}'. [1] https://code.launchpad.net/~siretart/+recipe/boxbackup-trunk > Thanks again for setting all this up, Achim You're welcome! I hope it results in better feedback for recent development versions. BTW, is what's the timeline for 0.11 and 0.12? -- Gruesse/greetings, Reinhard Tartler, KeyID 945348A4 From achim+box at qustodium.net Thu Mar 10 21:21:41 2011 From: achim+box at qustodium.net (Achim J. Latz) Date: Thu, 10 Mar 2011 22:21:41 +0100 Subject: [Box Backup] daily builds of trunk for ubuntu In-Reply-To: <87vczruk60.fsf@faui44a.informatik.uni-erlangen.de> References: <8762td12it.fsf@faui44a.informatik.uni-erlangen.de> <4D6677F2.2050408@qustodium.net> <87aahlkq9g.fsf@faui44a.informatik.uni-erlangen.de> <4D78C175.6000407@qustodium.net> <87vczruk60.fsf@faui44a.informatik.uni-erlangen.de> Message-ID: <4D7940E5.5060600@qustodium.net> Hello Reinhard: On 10/03/2011 15:25, Reinhard Tartler wrote: > Unfortunately, shell code is not possible. The available variables are > ${time}, and ${revno}. The documentation is here: > > https://help.launchpad.net/Packaging/SourceBuilds/BzrBuilder > > [..] > > This means the current version scheme is '0.12~~+trunk+r{revno}'. Well, at least adding the date will enable us to check roughly what branch was built via the timeline available in Trac [1] that I just discovered. > BTW, is what's the timeline for 0.11 and 0.12? Well, as I argued in my message "Development feedback & discussion for Box Backup 1.0?" from 14 January, I believe that the project is almost ready to obtain a "1.0" moniker. Chris has just received a new machine, perhaps he can share his perspective on the project? In general, I feel that contributions are pretty limited to a small group of people, which might be holding back the project. Best regards, Achim [1] -- Achim J. Latz, Qustodium Internet Security achim.latz at qustodium.net ? http://www.qustodium.net Data Encryption ? Backup Automatisation ? E-Mail Protection From ell2 at live.se Tue Mar 15 20:22:46 2011 From: ell2 at live.se (Leif Linderstam) Date: Tue, 15 Mar 2011 21:22:46 +0100 Subject: [Box Backup] Experiences with modification for unencrypted client-server communication Message-ID: This message is just to share my experiences with some tinkering with the BoxBackup code. The backup solution I am building for myself is a small box with a BoxBackup server on it. This box will be placed at a friend's house so that I get a true remote on-line backup solution. The box sets up a VPN tunnel to the file server machine back home, the one that needs to be backed up. So the network path between the BoxBackup client and server is already secured, which prompted me to try and add unencrypted communication to BoxBackup. The short story is that the approach I chose was quite easy to implement, but the solution uses a part of the code in BoxBackup in a way that it probably was not designed for. It worked for a small test with bbackupquery, but a longer running bbackupd got into trouble fairly soon. Ironing out these problems is beyond the amount of effort I want to put into this at the moment, so I decided to stick with the much more well-tested and reliable untouched code. Also, in the end the speed will probably be limited more by the Internet connection I have than by the extra encryption. That is about it. Those interested in the more technical details can read on. In the non-modified code the communication between client and server is done using the class SocketStreamTLS. This is derived from SocketStream, and among other things the member functions Read and Write are overridden. During connection setup and handshaking the class gets the client certificate, which is used to retrieve the name of the client and associate the connection with an account on the server. My approach was to keep this so that the changes in the code would be as small as possible. More or less I only made the Read and Write functions call the base class functions so that once the SSL link was set up, it would not really be used. Instead normal reads and writes would be performed on the underlying socket. The first problem was that SocketStream::Read seems to assume that the underlying socket is set to blocking mode, and the underlying socket used by SocketStreamTLS is set to non- blocking. I assumed that the original BoxBackup code never uses SocketStream directly, only SocketStreamTLS. (That assumption might have been totally wrong, I do not know.) So I changed SocketStream::Read to work with non-blocking sockets. It worked fairly well then, but after running bbackupd for a while it gave up on some timeout in the protocol code. At that point I gave up. Cheers, Leif From sjh_boxbackup at shic.co.uk Wed Mar 16 16:18:00 2011 From: sjh_boxbackup at shic.co.uk (Steve) Date: Wed, 16 Mar 2011 16:18:00 +0000 Subject: [Box Backup] The backup daemon on server reported an unknown error (backup-ok). Message-ID: <4D80E2B8.8080908@shic.co.uk> I've started using boxbackup - and have configured it to backup my home directory... about 14-15GB. It took about 20 hours (bbstored seeming to take most resources) and - then with a reassuring 14764mb in the backup directory... I get this cryptic email: -- Return-Path: X-Original-To: sjh at mailserver.example.com Delivered-To: sjh at mailserver.example.com Received: from localhost (localhost [127.0.0.1]) by gw.example.com (Postfix) with ESMTP id 6C03416C0BC1 for ; Wed, 16 Mar 2011 14:23:24 +0000 (GMT) Received: by gw.example.com (Postfix, from userid 0) id 4368F16C0A77; Wed, 16 Mar 2011 14:23:18 +0000 (GMT) Subject: BACKUP PROBLEM on host server (unknown) To: root at example.com Message-Id: <20110316142319.4368F16C0A77 at mailserver.example.com> Date: Wed, 16 Mar 2011 14:23:18 +0000 (GMT) From: root at example.com (root) The backup daemon on server reported an unknown error (backup-ok). ========================== FILES MAY NOT BE BACKED UP ========================== Please check the logs on server. -- If the backup really is OK, then that shouldn't be an error... IMHO... but I checked the logs - just in case. -- Mar 16 14:23:18 server bbstored client=0x00000001[29724]: NOTICE: Session finished for Client ID 0x00000001 Mar 16 14:23:18 server bbstored client=0x00000001[29724]: NOTICE: Connection statistics for BACKUP-00000001: IN=13785251581 OUT=14719428 TOTAL=13799971009 Mar 16 14:23:19 server bbackupd[29721]: NOTICE: Finished scan of local files Mar 16 14:23:19 server bbackupd[29721]: NOTICE: File statistics: total file size uploaded 18677462058, bytes already on server 0, encoded size 13712500888 Mar 16 14:23:19 server bbackupd[29721]: NOTICE: Beginning scan of local files Mar 16 14:23:19 server postfix/pickup[8338]: 4368F16C0A77: uid=0 from= Mar 16 14:23:19 server postfix/cleanup[8477]: 4368F16C0A77: message-id=<20110316142319.4368F16C0A77 at mailserver.example.com> Mar 16 14:23:19 server bbstored[28839]: NOTICE: Message from child process 8483: Incoming connection from 10.1.2.3 port 34133 Mar 16 14:23:19 server bbstored client=0x00000001[8483]: NOTICE: Login from Client ID 0x00000001 Read/Write Mar 16 14:23:19 server postfix/qmgr[25912]: 4368F16C0A77: from=, size=512, nrcpt=1 (queue active) Mar 16 14:23:24 server postfix/smtpd[8488]: connect from localhost[127.0.0.1] Mar 16 14:23:24 server postfix/smtpd[8488]: 6C03416C0BC1: client=localhost[127.0.0.1] Mar 16 14:23:24 server postfix/cleanup[8477]: 6C03416C0BC1: message-id=<20110316142319.4368F16C0A77 at mailserver.example.com> -- I don't see any errors reported in the log... Which makes me wonder - is this email evidence of a bug, or is it a glitch with my configuration? I'm using the default packages for boxbackup-client and boxbackup-server on Ubuntu Server 10.10. The executables report: Box Backup Store Server v0.11rc2+2502, (c) Ben Summers and contributors 2003-2008 Box Backup Client v0.11rc2+2502, (c) Ben Summers and contributors 2003-2008 I re-created my configuration from scratch... because I want continuous backup, as opposed to a snapshot approach. From chris at qwirx.com Wed Mar 16 21:37:09 2011 From: chris at qwirx.com (Chris Wilson) Date: Wed, 16 Mar 2011 21:37:09 +0000 (GMT) Subject: [Box Backup] Experiences with modification for unencrypted client-server communication In-Reply-To: References: Message-ID: Hi Leif, On Tue, 15 Mar 2011, Leif Linderstam wrote: > This message is just to share my experiences with some tinkering with > the BoxBackup code. The backup solution I am building for myself is a > small box with a BoxBackup server on it. This box will be placed at a > friend's house so that I get a true remote on-line backup solution. The > box sets up a VPN tunnel to the file server machine back home, the one > that needs to be backed up. So the network path between the BoxBackup > client and server is already secured, which prompted me to try and add > unencrypted communication to BoxBackup. Are you aware that removing the SSL layer also removes the ability for the server to verify the client's ID, which would allow any client to impersonate any other signed by the same CA? > The short story is that the approach I chose was quite easy to > implement, but the solution uses a part of the code in BoxBackup in a > way that it probably was not designed for. It worked for a small test > with bbackupquery, but a longer running bbackupd got into trouble fairly > soon. Ironing out these problems is beyond the amount of effort I want > to put into this at the moment, so I decided to stick with the much more > well-tested and reliable untouched code. Also, in the end the speed will > probably be limited more by the Internet connection I have than by the > extra encryption. Agreed about the speed issue. I'm not persuaded about the value of removing TLS encryption, versus the potential security risk. Perhaps configuring TLS to use a null cipher, and signing packets for authentication, might still be more secure than a completely unencrypted connection? > In the non-modified code the communication between client and server is > done using the class SocketStreamTLS. This is derived from SocketStream, > and among other things the member functions Read and Write are > overridden. During connection setup and handshaking the class gets the > client certificate, which is used to retrieve the name of the client and > associate the connection with an account on the server. My approach was > to keep this so that the changes in the code would be as small as > possible. More or less I only made the Read and Write functions call the > base class functions so that once the SSL link was set up, it would not > really be used. Instead normal reads and writes would be performed on > the underlying socket. Did you consider using a plain SocketStream instead of a SocketStreamTLS? > The first problem was that SocketStream::Read seems to assume that the > underlying socket is set to blocking mode, and the underlying socket > used by SocketStreamTLS is set to non- blocking. I assumed that the > original BoxBackup code never uses SocketStream directly, only > SocketStreamTLS. (That assumption might have been totally wrong, I do > not know.) I think we do have some tests for SocketStream (without TLS) in test/basicserver, although I could be wrong about that. I guess we don't use it in bbackupd or bbackupquery. > So I changed SocketStream::Read to work with non-blocking sockets. It > worked fairly well then, but after running bbackupd for a while it gave > up on some timeout in the protocol code. At that point I gave up. SocketStream should hide the details of whether the underlying socket is blocking or not, so this might be a bug in the applications using SocketStreamTLS, but probably not one that I'd be inclined to investigate for the above reasons. Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From chris at qwirx.com Wed Mar 16 21:49:42 2011 From: chris at qwirx.com (Chris Wilson) Date: Wed, 16 Mar 2011 21:49:42 +0000 (GMT) Subject: [Box Backup] The backup daemon on server reported an unknown error (backup-ok). In-Reply-To: <4D80E2B8.8080908@shic.co.uk> References: <4D80E2B8.8080908@shic.co.uk> Message-ID: Hi Steve, On Wed, 16 Mar 2011, Steve wrote: > The backup daemon on server reported an unknown error (backup-ok). Thanks for the bug report, I'll fix it as soon as e2fsck finishes checking my backup volume so that I can resume Box Backup development on my new desktop PC. Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From sjh at shic.co.uk Wed Mar 16 22:13:35 2011 From: sjh at shic.co.uk (Steve Haeck) Date: Wed, 16 Mar 2011 22:13:35 +0000 Subject: [Box Backup] The backup daemon on server reported an unknown error (backup-ok). In-Reply-To: References: <4D80E2B8.8080908@shic.co.uk> Message-ID: <4D81360F.8070807@shic.co.uk> On 16/03/2011 21:49, Chris Wilson wrote: > Hi Steve, > > On Wed, 16 Mar 2011, Steve wrote: > >> The backup daemon on server reported an unknown error (backup-ok). > > Thanks for the bug report, I'll fix it as soon as e2fsck finishes > checking my backup volume so that I can resume Box Backup development > on my new desktop PC. > > Cheers, Chris. Many thanks - I'd assumed I'd done something wrong... :) As an aside, congratulations on a product, that - on the whole - is exactly what I've been looking for for many years. Obviously, polish and performance improvements will be welcome... but, even as is, boxbackup represents a major step forwards for me. I wonder - is there a reason other than the pressure of time that leads to only supporting clients under Windows? Would it be hard to port bbstored to windows? Steve From chris at qwirx.com Thu Mar 17 21:32:42 2011 From: chris at qwirx.com (Chris Wilson) Date: Thu, 17 Mar 2011 21:32:42 +0000 (GMT) Subject: [Box Backup] The backup daemon on server reported an unknown error (backup-ok). In-Reply-To: <4D81360F.8070807@shic.co.uk> References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> Message-ID: Hi Steve, On Wed, 16 Mar 2011, Steve Haeck wrote: >>> The backup daemon on server reported an unknown error (backup-ok). >> >> Thanks for the bug report, I'll fix it as soon as e2fsck finishes >> checking my backup volume so that I can resume Box Backup development >> on my new desktop PC. > > Many thanks - I'd assumed I'd done something wrong... :) I've fixed the bbackupd-config script, but unless you want to run it again, using the new version in trunk, you'll need to edit the notify script that it generated (the one pointed to by NotifyScript in your config), find the following line: elif [ "\$1" = backup-start -o "\$1" = backup-finish ]; then and change it to: elif [ "\$1" = backup-start -o "\$1" = backup-finish \ -o "\$1" = backup-ok ]; then > As an aside, congratulations on a product, that - on the whole - is > exactly what I've been looking for for many years. Obviously, polish > and performance improvements will be welcome... but, even as is, > boxbackup represents a major step forwards for me. Thanks, but I can't take much credit, I inherited most of this code from the inimitable Ben Summers and other contributors, and I just keep the code alive as best I can in my spare time :) > I wonder - is there a reason other than the pressure of time that leads > to only supporting clients under Windows? Would it be hard to port > bbstored to windows? There is an unsupported Windows port of bbstored, which is built as part of running the unit tests. It does have serious limitations; because bbstored relies on the fork() model on Unix, which is not possible on Windows, the server is limited to one concurrent connection, and of course is not well tested. But you're welcome to try it. Why exactly do you want to run bbstored on Windows? Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From sjh_boxbackup at shic.co.uk Fri Mar 18 11:14:59 2011 From: sjh_boxbackup at shic.co.uk (Steve) Date: Fri, 18 Mar 2011 11:14:59 +0000 Subject: [Box Backup] The backup daemon on server reported an unknown error (backup-ok). In-Reply-To: References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> Message-ID: <4D833EB3.4020002@shic.co.uk> On 17/03/2011 21:32, Chris Wilson wrote: > I've fixed the bbackupd-config script, but unless you want to run it > again, using the new version in trunk, you'll need to edit the notify > script that it generated (the one pointed to by NotifyScript in your > config), find the following line: > > elif [ "\$1" = backup-start -o "\$1" = backup-finish ]; then > > and change it to: > > elif [ "\$1" = backup-start -o "\$1" = backup-finish \ > -o "\$1" = backup-ok ]; then I've opted to hack my notification script - without the escaped '$' characters, of course... I'm happy to wait for the next release to be able to cleanly configure again. I'd hesitated to do this hack previously as I'd wondered how 'backup-ok' differed from 'backup-finish' and whether it really did document something I should care about. >> As an aside, congratulations on a product, that - on the whole - is >> exactly what I've been looking for for many years. Obviously, polish >> and performance improvements will be welcome... but, even as is, >> boxbackup represents a major step forwards for me. > > Thanks, but I can't take much credit, I inherited most of this code > from the inimitable Ben Summers and other contributors, and I just > keep the code alive as best I can in my spare time :) It's appreciated. :) >> I wonder - is there a reason other than the pressure of time that >> leads to only supporting clients under Windows? Would it be hard to >> port bbstored to windows? > > There is an unsupported Windows port of bbstored, which is built as > part of running the unit tests. It does have serious limitations; > because bbstored relies on the fork() model on Unix, which is not > possible on Windows, the server is limited to one concurrent > connection, and of course is not well tested. But you're welcome to > try it. Why exactly do you want to run bbstored on Windows? Hmmm... My reasons, first: I've two almost-always-on PCs - one is a (moderately inaccessible) Linux server - with Raid-1 - the other a Windows desktop (on which I occasionally run virtualbox in seemless mode for any linux-only UI work.) The two boxes are separated by an Ethernet run - and this means I have maximum high-bandwidth-connected separation between these two physical locations. If bbstored uses different hardware, in a different location, it would mitigate more risks. An architecture dependent on fork (while I only anticipate a single client) suggests that work would be needed... and I've just realised that I'd end-up storing the files on NTFS rather than ext* - though I'm not sure what implications that has for me. Perhaps I should look at lightweight ways to expose block devices to my linux server over my LAN - to get the same advantage of physical separation of backup from live data... though I'm wary of this, too - since I want maximum reliability. From chris at qwirx.com Fri Mar 18 12:55:25 2011 From: chris at qwirx.com (Chris Wilson) Date: Fri, 18 Mar 2011 12:55:25 +0000 (GMT) Subject: [Box Backup] The backup daemon on server reported an unknown error (backup-ok). In-Reply-To: <4D833EB3.4020002@shic.co.uk> References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> <4D833EB3.4020002@shic.co.uk> Message-ID: Hi Steve, On Fri, 18 Mar 2011, Steve wrote: > I've opted to hack my notification script - without the escaped '$' > characters, of course... I'm happy to wait for the next release to be > able to cleanly configure again. I'd hesitated to do this hack > previously as I'd wondered how 'backup-ok' differed from 'backup-finish' > and whether it really did document something I should care about. backup-finish doesn't indicate the status, but could be useful for cleanup tasks or sending reports regardless of the status. > An architecture dependent on fork (while I only anticipate a single > client) If you only anticipate a single (simultaneous) client then there should be no problems. Even two clients might be OK, as one will just wait (maybe failing to connect to the server and retrying) until the other has finished and pauses. > suggests that work would be needed... and I've just realised > that I'd end-up storing the files on NTFS rather than ext* - though I'm > not sure what implications that has for me. You can use the ext2 IFS on Windows if NTFS bother you, but more importantly, Windows forbids renaming over existing files, which means that the atomic file replacement guarantee of which Ben was very proud, and which probably underlies various assumptions about data security elsewhere in Box Backup, simply isn't possible. You could also run Box Backup in a virtual machine or coLinux on your Windows box? Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \ _/_/_/_//_/___/ | Stop nuclear war http://www.nuclearrisk.org | From ell2 at live.se Fri Mar 18 21:56:08 2011 From: ell2 at live.se (Leif Linderstam) Date: Fri, 18 Mar 2011 22:56:08 +0100 Subject: [Box Backup] Experiences with modification for unencrypted client-server communication Message-ID: Hi Chris, > Are you aware that removing the SSL layer also removes the ability for the > server to verify the client's ID, which would allow any client to > impersonate any other signed by the same CA? Yes, I am aware of that, but this is a home network and all machines on it are trusted not to behave that way. Through the VPN tunnel the remote box becomes part of my home network and is almost invisible to the rest of the world. > Agreed about the speed issue. I'm not persuaded about the value of > removing TLS encryption, versus the potential security risk. Perhaps > configuring TLS to use a null cipher, and signing packets for > authentication, might still be more secure than a completely unencrypted > connection? I would not use non-encrypted communication over the Internet, or any other not entirely trusted network for that matter. Nevertheless I also have come to the conclusion that there is not enough value in being able to remove encryption. Mostly because of the added complexity. > Did you consider using a plain SocketStream instead of a SocketStreamTLS? That would be the proper way of doing it, yes. But I was looking for the quickest way to turn off encryption for the main bulk of data. Additionally the server logic uses the certificate that I guess it receives from the client to extract the "name" of the client, which in turn is used to associate the session with the right account. Using SocketStream directly would have meant adding a corresponding handshake as well. > SocketStream should hide the details of whether the underlying socket is > blocking or not, so this might be a bug in the applications using > SocketStreamTLS, but probably not one that I'd be inclined to investigate > for the above reasons. No, of course not. Do not waste any time on this. I just wrote the mail to share my experiences if someone was interested. If my memory serves me someone has previously asked about this sort of thing but I cannot find it at the moment. Cheers, Leif From a.chapellon at horoa.net Thu Mar 24 22:46:18 2011 From: a.chapellon at horoa.net (Alexandre Chapellon) Date: Thu, 24 Mar 2011 23:46:18 +0100 Subject: [Box Backup] test errors Message-ID: <1301006778.2595.97.camel@elronde.middle.earth> Hi everyone, Am new to backupbox and am very interrested in this project. I then decided to test it on my fresh debian squeeze. I got the 0.11rc_ tarball and did the usual tricks (./configure && make) When doing `make test` the system works for a while (I have to say i didn't read the whole output :p) and finally gives me this sumarry: common: PASSED crypto: PASSED ... httpserver: FAILED: 1 tests failed (first at testhttpserver.cpp:214) I have checked the common causes to make sure it was ok... and it was: - enough free space - no firewall - no running bb instances - I guess my virtual machine is fast enough to complete test (only a guess: Athlon X2 3ghz / 512MB / Xen-HVM with LVM sata disks) Before goiing further I'd like to know if it's failing this test is a problem? Does anyone know why it would fails? Additionnally I do not run make test as root but as a normal user. Best regards. -- horoa: la voie est libre -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 490 bytes Desc: This is a digitally signed message part URL: From chris at qwirx.com Thu Mar 24 23:39:15 2011 From: chris at qwirx.com (Chris Wilson) Date: Thu, 24 Mar 2011 23:39:15 +0000 (GMT) Subject: [Box Backup] test errors In-Reply-To: <1301006778.2595.97.camel@elronde.middle.earth> References: <1301006778.2595.97.camel@elronde.middle.earth> Message-ID: Hi Alexandre, On Thu, 24 Mar 2011, Alexandre Chapellon wrote: > Am new to backupbox and am very interrested in this project. I then > decided to test it on my fresh debian squeeze. I got the 0.11rc_ tarball > and did the usual tricks (./configure && make) When doing `make test` > the system works for a while (I have to say i didn't read the whole > output :p) and finally gives me this sumarry: > > common: PASSED > crypto: PASSED > ... > httpserver: FAILED: 1 tests failed (first at testhttpserver.cpp:214) > > I have checked the common causes to make sure it was ok... and it was: > - enough free space > - no firewall > - no running bb instances > - I guess my virtual machine is fast enough to complete test (only a guess: Athlon X2 3ghz / 512MB / Xen-HVM with LVM sata disks) > > Before goiing further I'd like to know if it's failing this test is a problem? Does anyone know why it would fails? It often fails because some Perl modules that it requires are not installed. Please consult the test output logs for more details. httpserver is not used at all in current versions of Box Backup, so it's safe to completely ignore failure of this test. > Additionnally I do not run make test as root but as a normal user. Good idea :) > horoa: la voie est libre I don't understand this, and google translate doesn't help much :) What does it mean? Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From a.chapellon at horoa.net Thu Mar 24 23:48:02 2011 From: a.chapellon at horoa.net (Alexandre Chapellon) Date: Fri, 25 Mar 2011 00:48:02 +0100 Subject: [Box Backup] test errors In-Reply-To: References: <1301006778.2595.97.camel@elronde.middle.earth> Message-ID: <1301010482.2595.103.camel@elronde.middle.earth> Le jeudi 24 mars 2011 ? 23:39 +0000, Chris Wilson a ?crit : > Hi Alexandre, > > On Thu, 24 Mar 2011, Alexandre Chapellon wrote: > > > Am new to backupbox and am very interrested in this project. I then > > decided to test it on my fresh debian squeeze. I got the 0.11rc_ tarball > > and did the usual tricks (./configure && make) When doing `make test` > > the system works for a while (I have to say i didn't read the whole > > output :p) and finally gives me this sumarry: > > > > common: PASSED > > crypto: PASSED > > ... > > httpserver: FAILED: 1 tests failed (first at testhttpserver.cpp:214) > > > > I have checked the common causes to make sure it was ok... and it was: > > - enough free space > > - no firewall > > - no running bb instances > > - I guess my virtual machine is fast enough to complete test (only a guess: Athlon X2 3ghz / 512MB / Xen-HVM with LVM sata disks) > > > > Before goiing further I'd like to know if it's failing this test is a problem? Does anyone know why it would fails? > > It often fails because some Perl modules that it requires are not > installed. Please consult the test output logs for more details. > httpserver is not used at all in current versions of Box Backup, so it's > safe to completely ignore failure of this test. If I understand correctly it's only for future use.. right? > > > Additionnally I do not run make test as root but as a normal user. > > Good idea :) > > > horoa: la voie est libre > > I don't understand this, and google translate doesn't help much :) What > does it mean? > This is my signature sorry, maybe it's not very appropiate here. horoa is the name of company (a word for 'to give' in tahitian) and "la voie est libre" is a french idiom meaning "the way is clear", furthermore 'libre' means free as in free speech and is of course a reference to free softwares... This is nit related to this mailing list at all but I wanted to make it clear :) > Cheers, Chris. -- horoa: la voie est libre -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 490 bytes Desc: This is a digitally signed message part URL: From chris at qwirx.com Thu Mar 24 23:55:13 2011 From: chris at qwirx.com (Chris Wilson) Date: Thu, 24 Mar 2011 23:55:13 +0000 (GMT) Subject: [Box Backup] test errors In-Reply-To: <1301010482.2595.103.camel@elronde.middle.earth> References: <1301006778.2595.97.camel@elronde.middle.earth> <1301010482.2595.103.camel@elronde.middle.earth> Message-ID: Hi Alexandre, On Fri, 25 Mar 2011, Alexandre Chapellon wrote: > > > httpserver: FAILED: 1 tests failed (first at testhttpserver.cpp:214) > > > httpserver is not used at all in current versions of Box Backup, so > > it's safe to completely ignore failure of this test. > > If I understand correctly it's only for future use.. right? Yes, that's correct. Please note that you are sending HTML emails which make it difficult for me to quote your replies properly, and many people consider them inappropriate on public Unixish mailing lists. > > horoa: la voie est libre > > This is my signature sorry, maybe it's not very appropiate here. horoa > is the name of company (a word for 'to give' in tahitian) and "la voie > est libre" is a french idiom meaning "the way is clear", furthermore > 'libre' means free as in free speech and is of course a reference to > free softwares... This is nit related to this mailing list at all but I > wanted to make it clear :) Thanks for the clarification :) I was thinking of "empty train station platforms" which didn't make much sense :) Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From a.chapellon at horoa.net Fri Mar 25 09:48:04 2011 From: a.chapellon at horoa.net (Alexandre Chapellon) Date: Fri, 25 Mar 2011 10:48:04 +0100 Subject: [Box Backup] default config dir Message-ID: <1301046484.3299.21.camel@elronde.middle.earth> Hello, This morning, trying to go ahead with boxbackup I noticed an error while trying to create an account on the server. sudo bbstoreaccounts create 1 0 2000 2048 ERROR: FileHandleGuard: failed to open file '/etc/boxbackup/bbstored.conf': No such file or directory (2) TRACE: Obtained 6 stack frames. TRACE: Stack frame 0: bbstoreaccounts(DumpStackBacktrace()+0x26) [0x48f176] TRACE: Stack frame 1: bbstoreaccounts(FileHandleGuard<0, 438>::FileHandleGuard(std::string const&)+0x38d) [0x41f29d] TRACE: Stack frame 2: bbstoreaccounts(Configuration::LoadAndVerify(std::string const&, ConfigurationVerify const*, std::string&)+0x54) [0x47b124] TRACE: Stack frame 3: bbstoreaccounts(main+0x12a) [0x41a8aa] TRACE: Stack frame 4: /lib/libc.so.6(__libc_start_main+0xfd) [0x7f9334479c4d] TRACE: Stack frame 5: bbstoreaccounts() [0x416129] WARNING: Exception thrown: CommonException(OSFileOpenError) at ../../lib/common/Guards.h(102) Exception: Common OSFileOpenError (Can't open a file -- attempted to load a non-existant config file or bad file referenced within?) (1/2) Indeed the /etc/boxbackup directory doesn't exist, because I didn't created it. But the online wiki shows /etc/box as the default config-dir. Furthermore the commented source code in ./lib/server/Daemon.cpp:449 seems to tell that /etc/boxbackup is deprecated in favour of /etc/box. It seems to me that parts of the code still uses /etc/boxbackup. Is it something made on purpose for some backward compat or is it just the current (to be changed) state of the rc8? -- horoa: la voie est libre -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 490 bytes Desc: This is a digitally signed message part URL: From chris at qwirx.com Fri Mar 25 11:57:42 2011 From: chris at qwirx.com (Chris Wilson) Date: Fri, 25 Mar 2011 11:57:42 +0000 (GMT) Subject: [Box Backup] default config dir In-Reply-To: <1301046484.3299.21.camel@elronde.middle.earth> References: <1301046484.3299.21.camel@elronde.middle.earth> Message-ID: Hi Alexandre, On Fri, 25 Mar 2011, Alexandre Chapellon wrote: > ERROR: FileHandleGuard: failed to open file > '/etc/boxbackup/bbstored.conf': No such file or directory (2) > > Indeed the /etc/boxbackup directory doesn't exist, because I didn't > created it. > But the online wiki shows /etc/box as the default config-dir. That's out of date, where? > Furthermore the commented source code in ./lib/server/Daemon.cpp:449 > seems to tell that /etc/boxbackup is deprecated in favour of /etc/box. Other way around. > It seems to me that parts of the code still uses /etc/boxbackup. Is it > something made on purpose for some backward compat or is it just the > current (to be changed) state of the rc8? Please use /etc/boxbackup. Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \ _/_/_/_//_/___/ | Stop nuclear war http://www.nuclearrisk.org | From a.chapellon at horoa.net Fri Mar 25 12:03:04 2011 From: a.chapellon at horoa.net (Alexandre Chapellon) Date: Fri, 25 Mar 2011 13:03:04 +0100 Subject: [Box Backup] default config dir In-Reply-To: References: <1301046484.3299.21.camel@elronde.middle.earth> Message-ID: <1301054584.3299.29.camel@elronde.middle.earth> Le vendredi 25 mars 2011 ? 11:57 +0000, Chris Wilson a ?crit : > Hi Alexandre, > > On Fri, 25 Mar 2011, Alexandre Chapellon wrote: > > > ERROR: FileHandleGuard: failed to open file > > '/etc/boxbackup/bbstored.conf': No such file or directory (2) > > > > Indeed the /etc/boxbackup directory doesn't exist, because I didn't > > created it. > > But the online wiki shows /etc/box as the default config-dir. > > That's out of date, where? http://www.boxbackup.org/client.html http://www.boxbackup.org/server.html > > > Furthermore the commented source code in ./lib/server/Daemon.cpp:449 > > seems to tell that /etc/boxbackup is deprecated in favour of /etc/box. > > Other way around. > > > It seems to me that parts of the code still uses /etc/boxbackup. Is it > > something made on purpose for some backward compat or is it just the > > current (to be changed) state of the rc8? > > Please use /etc/boxbackup. okay > > Cheers, Chris. -- horoa: la voie est libre -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 490 bytes Desc: This is a digitally signed message part URL: From pjalajas at gigalock.com Fri Mar 25 17:48:36 2011 From: pjalajas at gigalock.com (Peter Jalajas, GigaLock Backup Services) Date: Fri, 25 Mar 2011 13:48:36 -0400 Subject: [Box Backup] default config dir In-Reply-To: <1301054584.3299.29.camel@elronde.middle.earth> References: <1301046484.3299.21.camel@elronde.middle.earth> <1301054584.3299.29.camel@elronde.middle.earth> Message-ID: Hi Alexandre, FWIW, when I've encountered that problem: >> > '/etc/boxbackup/bbstored.conf': No such file or directory (2) I simply made a symlink, something like so: sudo ln -s /etc/box /etc/boxbackup Hope that helps a little, Pete On Fri, Mar 25, 2011 at 8:03 AM, Alexandre Chapellon wrote: > Le vendredi 25 mars 2011 ? 11:57 +0000, Chris Wilson a ?crit : >> Hi Alexandre, >> >> On Fri, 25 Mar 2011, Alexandre Chapellon wrote: >> >> > ERROR: ? FileHandleGuard: failed to open file >> > >> > Indeed the /etc/boxbackup directory doesn't exist, because I didn't >> > created it. >> > But the online wiki shows /etc/box as the default config-dir. >> >> That's out of date, where? > > http://www.boxbackup.org/client.html > http://www.boxbackup.org/server.html > >> >> > Furthermore the commented source code in ./lib/server/Daemon.cpp:449 >> > seems to tell that /etc/boxbackup is deprecated in favour of /etc/box. >> >> Other way around. >> >> > It seems to me that parts of the code still uses /etc/boxbackup. Is it >> > something made on purpose for some backward compat or is it just the >> > current (to be changed) state of the rc8? >> >> Please use /etc/boxbackup. > okay > >> >> Cheers, Chris. > > -- > horoa: la voie est libre > > _______________________________________________ > Boxbackup mailing list > Boxbackup at boxbackup.org > http://lists.boxbackup.org/cgi-bin/mailman/listinfo/boxbackup > > From chris at qwirx.com Fri Mar 25 23:45:04 2011 From: chris at qwirx.com (Chris Wilson) Date: Fri, 25 Mar 2011 23:45:04 +0000 (GMT) Subject: [Box Backup] default config dir In-Reply-To: <1301054584.3299.29.camel@elronde.middle.earth> References: <1301046484.3299.21.camel@elronde.middle.earth> <1301054584.3299.29.camel@elronde.middle.earth> Message-ID: Hi all, On Fri, 25 Mar 2011, Alexandre Chapellon wrote: >>> Indeed the /etc/boxbackup directory doesn't exist, because I didn't >>> created it. But the online wiki shows /etc/box as the default >>> config-dir. >> >> That's out of date, where? > > http://www.boxbackup.org/client.html > http://www.boxbackup.org/server.html OK, that's not the wiki, it's the static site which is more out of date. I suggested a while ago (perhaps a couple of years) completely replacing the old static website with the Trac wiki. I didn't hear any objections at the time. Does anyone object now? If nobody contacts me, I may well go ahead and replace the old site without further warning. Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From chris at qwirx.com Fri Mar 25 23:46:56 2011 From: chris at qwirx.com (Chris Wilson) Date: Fri, 25 Mar 2011 23:46:56 +0000 (GMT) Subject: [Box Backup] Norton blocking boxbackup.org? In-Reply-To: <20110109110730.GJ21506@netinertia.co.uk> References: <96da9c7754da17451aac43d2221d82f9@kaufmanfamily.net> <9E94BF29-5C0D-4802-AD53-95166EBED80C@netinertia.co.uk> <20110109110730.GJ21506@netinertia.co.uk> Message-ID: Hi James, On Sun, 9 Jan 2011, James O'Gorman wrote: > On Sun, Jan 09, 2011 at 01:50:50AM +0000, Chris Wilson wrote: >>>> I proposed boxbackup to a colleague recently, but when he hit the web >>>> site for more info he was blocked by Norton SafeWeb. Apparently their >>>> scanner thinks it has found an old Windows XP key logger in one of the >>>> boxbackup tarballs - more details at this link: >>>> >>>> http://safeweb.norton.com/report/show?url=http:%2F%2Fwww.boxbackup.org%2Ftrac%2Fwiki >>>> >>>> There's also a place to click if you're the site owner, presumably to >>>> clear up the warning. Needless to say my colleague was a little put off >>>> by the warning. >>> >>> Thanks for letting us know. I'll register myself as the site owner and >>> see if I can get that re-evaluated. >> >> Did you get anywhere with this? > > I did, and they removed the site report, but it seems they've reinstated > it again. I suppose their heuristics are matching some string on that > file that causes them to think there's something bad there. I found out what it is. The tarballs contain a program called RemoteControl.exe in contrib/windows/installer/tools, which Norton classifies as a trojan. I'm not even convinced that we have the right to redistribute any of the programs in that directory. Would anyone object if I remove them entirely? Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From pjalajas at gigalock.com Sat Mar 26 03:39:47 2011 From: pjalajas at gigalock.com (Peter Jalajas, GigaLock Backup Services) Date: Fri, 25 Mar 2011 23:39:47 -0400 Subject: [Box Backup] Norton blocking boxbackup.org? In-Reply-To: References: <96da9c7754da17451aac43d2221d82f9@kaufmanfamily.net> <9E94BF29-5C0D-4802-AD53-95166EBED80C@netinertia.co.uk> <20110109110730.GJ21506@netinertia.co.uk> Message-ID: Hi Chris, On Fri, Mar 25, 2011 at 7:46 PM, Chris Wilson wrote: > I found out what it is. The tarballs contain a program called > RemoteControl.exe in contrib/windows/installer/tools, which Norton > classifies as a trojan. > > I'm not even convinced that we have the right to redistribute any of the > programs in that directory. Would anyone object if I remove them entirely? I created that folder and that .exe. RemoteControl.exe and grc.exe are specific to my system, are useless to anyone else, and should be deleted immediately and permanently. vshadow.exe is a Microsoft file and should be specifically deleted immediately as well. All other .exe, .dll, and .zip files in that directory and below should be deleted (notably also including those in the .../tools/redistributed/ directory) until someone can authoritatively review their licenses. They are essentially all part of the 7zip project, http://www.7-zip.org/license.txt (LGPL + additional required unRAR restriction). I used 7zip as part of the installer to unzip a password-encrypted package of boxbackup keys. I have long-since stopped using account-specific installers. Sorry for any inconvenience. Thanks, Pete From ell2 at live.se Sat Mar 26 13:19:42 2011 From: ell2 at live.se (Leif Linderstam) Date: Sat, 26 Mar 2011 14:19:42 +0100 Subject: [Box Backup] default config dir Message-ID: Hi Chris, >>> That's out of date, where? >> >> http://www.boxbackup.org/client.html >> http://www.boxbackup.org/server.html > > OK, that's not the wiki, it's the static site which is more out of date. Actually the wiki says the same. For instance: https://www.boxbackup.org/trac/wiki/ConfiguringAServer has the example: /usr/local/bin/bbstored-config /etc/box hostname _bbstored Cheers, Leif From chris at qwirx.com Sat Mar 26 14:45:45 2011 From: chris at qwirx.com (Chris Wilson) Date: Sat, 26 Mar 2011 14:45:45 +0000 (GMT) Subject: [Box Backup] default config dir In-Reply-To: References: Message-ID: Hi Leif, On Sat, 26 Mar 2011, Leif Linderstam wrote: >> OK, that's not the wiki, it's the static site which is more out of >> date. > > Actually the wiki says the same. For instance: > > https://www.boxbackup.org/trac/wiki/ConfiguringAServer > > has the example: > > /usr/local/bin/bbstored-config /etc/box hostname _bbstored Thanks for pointing that out, fixed now. Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From ell2 at live.se Sat Mar 26 15:37:35 2011 From: ell2 at live.se (Leif Linderstam) Date: Sat, 26 Mar 2011 16:37:35 +0100 Subject: [Box Backup] default config dir In-Reply-To: References: , Message-ID: Hi Chris, By the way, is the default path configurable when running the configure script? The installation path is configurable, but I didn't see anything about the config file path. This reminds me about another difference between the source and the wiki. By default the configure script sets the installation path for the binaries to ? /usr/local/sbin and the wiki uses ? /usr/local/bin at least in the example I gave previously from the server config page. ? Cheers, Leif From james at netinertia.co.uk Sat Mar 26 17:13:25 2011 From: james at netinertia.co.uk (James O'Gorman) Date: Sat, 26 Mar 2011 17:13:25 +0000 Subject: [Box Backup] default config dir In-Reply-To: References: Message-ID: <20110326171325.GO45156@netinertia.co.uk> On Sat, Mar 26, 2011 at 04:37:35PM +0100, Leif Linderstam wrote: > > Hi Chris, > > By the way, is the default path configurable when running the configure > script? The installation path is configurable, but I didn't see anything > about the config file path. Yes, I made a change to configure.in a while back to allow changing sysconfdir. When you run './configure --sysconfdir=/foo' it updates the header file so all of the Box utilities reference it. ./configure --help gives you all options. > This reminds me about another difference between the source and the wiki. > By default the configure script sets the installation path for the > binaries to > > ? /usr/local/sbin > > and the wiki uses > > ? /usr/local/bin > > at least in the example I gave previously from the server config page. That would be my bad. I made the change to put everything in $PREFIX/sbin instead of $PREFIX/bin because, generally speaking, non-root users would not use the programs. I'll see about updating the wiki for that. James From jwark at eastlink.ca Sat Mar 26 18:54:56 2011 From: jwark at eastlink.ca (Jack Warkentin) Date: Sat, 26 Mar 2011 15:54:56 -0300 Subject: [Box Backup] Replacing static website (was Re: default config dir) In-Reply-To: References: <1301046484.3299.21.camel@elronde.middle.earth> <1301054584.3299.29.camel@elronde.middle.earth> Message-ID: <4D8E3680.1030600@eastlink.ca> Hi Chris As has been identified by several messages in this thread, there are discrepancies between the old static website and the wiki. It would appear that sometimes the "old" is correct and sometimes the wiki. What is *really* needed is for someone to go through both and arrive at an amalgamated site that is as correct as possible. Only someone very familiar with the details of both code and configuration would have the knowledge to do that. That said, I personally have found the "old" more easy to navigate to the information I need. regards Jack Chris Wilson wrote: > Hi all, > > On Fri, 25 Mar 2011, Alexandre Chapellon wrote: > >>>> Indeed the /etc/boxbackup directory doesn't exist, because I didn't >>>> created it. But the online wiki shows /etc/box as the default >>>> config-dir. >>> >>> That's out of date, where? >> >> http://www.boxbackup.org/client.html http://www.boxbackup.org/server.html > > OK, that's not the wiki, it's the static site which is more out of date. > > I suggested a while ago (perhaps a couple of years) completely replacing > the old static website with the Trac wiki. I didn't hear any objections > at the time. Does anyone object now? If nobody contacts me, I may well > go ahead and replace the old site without further warning. > > Cheers, Chris. -- Jack Warkentin, phone 902-404-0457, email jwark at eastlink.ca 39 Inverness Avenue, Halifax, Nova Scotia, Canada, B3P 1X6 From chris at qwirx.com Sat Mar 26 19:41:23 2011 From: chris at qwirx.com (Chris Wilson) Date: Sat, 26 Mar 2011 19:41:23 +0000 (GMT) Subject: [Box Backup] Replacing static website (was Re: default config dir) In-Reply-To: <4D8E3680.1030600@eastlink.ca> References: <1301046484.3299.21.camel@elronde.middle.earth> <1301054584.3299.29.camel@elronde.middle.earth> <4D8E3680.1030600@eastlink.ca> Message-ID: Hi Jack and all, On Sat, 26 Mar 2011, Jack Warkentin wrote: > As has been identified by several messages in this thread, there are > discrepancies between the old static website and the wiki. It would > appear that sometimes the "old" is correct and sometimes the wiki. > > What is *really* needed is for someone to go through both and arrive at > an amalgamated site that is as correct as possible. Only someone very > familiar with the details of both code and configuration would have the > knowledge to do that. I think this was already done, many years ago. I'm very curious as to what information might be more up-to-date on the old static site, given the difficulty of updating it. Can anyone point me to something? > That said, I personally have found the "old" more easy to navigate to > the information I need. I think the old site has far less information on it, which may be one reason why it's easier to navigate. I'd also be interested to hear what kind of information people most often use on either site, to help determine what should have higher prominence on the trac wiki, to make it easier to navigate. Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From bjarne at mail2net.dk Sun Mar 27 21:02:52 2011 From: bjarne at mail2net.dk (Bjarne Carlsen) Date: Sun, 27 Mar 2011 22:02:52 +0200 Subject: [Box Backup] Replacing static website (was Re: default config dir) In-Reply-To: References: <1301046484.3299.21.camel@elronde.middle.earth> <1301054584.3299.29.camel@elronde.middle.earth> <4D8E3680.1030600@eastlink.ca> Message-ID: <4D8F97EC.2020205@mail2net.dk> On 26-03-2011 20:41, Chris Wilson wrote: > Hi Jack and all, > > > I think the old site has far less information on it, which may be one > reason why it's easier to navigate. I'd also be interested to hear > what kind of information people most often use on either site, to help > determine what should have higher prominence on the trac wiki, to make > it easier to navigate. > > Cheers, Chris. Hi Chris, Since I (sadly) don't have the time to dabble in the code, but do use the wiki frequently to read up on especially admin tasks, I vote for those as high prominence topics together with error messages and -handling. Cheers, Bjarne From ell2 at live.se Mon Mar 28 11:44:43 2011 From: ell2 at live.se (Leif Linderstam) Date: Mon, 28 Mar 2011 12:44:43 +0200 Subject: [Box Backup] default config dir In-Reply-To: <20110326171325.GO45156@netinertia.co.uk> References: , , , <20110326171325.GO45156@netinertia.co.uk> Message-ID: Hi James, > Yes, I made a change to configure.in a while back to allow changing > sysconfdir. When you run './configure --sysconfdir=/foo' it updates the > header file so all of the Box utilities reference it. > > ./configure --help gives you all options. Thank you for the info. I think I looked at that option but never tried it. What through me off was that the default value according to the help is PREFIX/etc, where PREFIX is /usr/local. The default however clearly is /etc/boxbackup so I guessed that the --sysconfdir was something else and probably not used at all. Cheers, Leif From ell2 at live.se Mon Mar 28 11:56:20 2011 From: ell2 at live.se (Leif Linderstam) Date: Mon, 28 Mar 2011 12:56:20 +0200 Subject: [Box Backup] default config dir In-Reply-To: References: , , , , , , <20110326171325.GO45156@netinertia.co.uk>, Message-ID: > it. What through me off was that the default value according to the help Oups, wrong word. "through" should have been "threw" of course. Cheers, Leif From james at netinertia.co.uk Mon Mar 28 19:02:45 2011 From: james at netinertia.co.uk (James O'Gorman) Date: Mon, 28 Mar 2011 19:02:45 +0100 Subject: [Box Backup] default config dir In-Reply-To: References: <20110326171325.GO45156@netinertia.co.uk> Message-ID: <20110328180244.GQ45156@netinertia.co.uk> Hi Leif, On Mon, Mar 28, 2011 at 12:44:43PM +0200, Leif Linderstam wrote: > Thank you for the info. I think I looked at that option but never tried > it. What through me off was that the default value according to the help > is PREFIX/etc, where PREFIX is /usr/local. The default however clearly is > /etc/boxbackup so I guessed that the --sysconfdir was something else and > probably not used at all. We put a bit of a fudge in configure.ac to force it to use /etc as the default. There's no way of making configure reflect this because it (quite rightly) thinks that everything should go under PREFIX, but a decision was made within the project to break this rule. You can still use ./configure --sysconfdir to change it, though. I use this for the FreeBSD port (in fact I think I had to fix it :-). James From sjh_boxbackup at shic.co.uk Wed Mar 30 16:35:33 2011 From: sjh_boxbackup at shic.co.uk (Steve) Date: Wed, 30 Mar 2011 16:35:33 +0100 Subject: [Box Backup] Reboot server.... "unknown error (backup-error)" In-Reply-To: <4D833EB3.4020002@shic.co.uk> References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> <4D833EB3.4020002@shic.co.uk> Message-ID: <4D934DC5.2020406@shic.co.uk> I'm using the default scripts to run boxbackup client and server from Ubuntu Server 10.10, and - as far as I can tell - the client is starting before the server... and is failing to connect - hence generating an error message. Is the client intended to delay startup until the server has started where both are installed on the same host? If not, would it be easy to implement it as a feature (the fewer errors in log files that need to be manually verified, the better.) From jwark at eastlink.ca Wed Mar 30 19:30:31 2011 From: jwark at eastlink.ca (Jack Warkentin) Date: Wed, 30 Mar 2011 15:30:31 -0300 Subject: [Box Backup] Reboot server.... "unknown error (backup-error)" In-Reply-To: <4D934DC5.2020406@shic.co.uk> References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> <4D833EB3.4020002@shic.co.uk> <4D934DC5.2020406@shic.co.uk> Message-ID: <4D9376C7.5040901@eastlink.ca> Hi Steve I had exactly this problem with my Debian system (from which Ubuntu is derived) and solved it in the following way. This must be done as root. # cd /etc # ls -d rc* rc0.d/ rc1.d/ rc2.d/ rc3.d/ rc4.d/ rc5.d/ rc6.d/ rc.local* rcS.d/ The rc* directories control the starting and stopping of various services at each of the various runlevels. In general, rc?.d does this for runlevel ?. This is done by creating symbolic links to the scripts in /etc/init.d. Here are the names of the symbolic links for boxbackup after fixing them. # find rc* -type l -iname '*boxbackup*' rc0.d/K21boxbackup-server rc0.d/K20boxbackup-client rc1.d/K21boxbackup-server rc1.d/K20boxbackup-client rc2.d/S21boxbackup-client rc2.d/S20boxbackup-server rc3.d/S21boxbackup-client rc3.d/S20boxbackup-server rc4.d/S21boxbackup-client rc4.d/S20boxbackup-server rc5.d/S21boxbackup-client rc5.d/S20boxbackup-server rc6.d/K21boxbackup-server rc6.d/K20boxbackup-client Here is what they actually look like. # find rc* -type l -iname '*boxbackup*' -exec ls -al \{\} \; lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc0.d/K21boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc0.d/K20boxbackup-client -> ../init.d/boxbackup-client lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc1.d/K21boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc1.d/K20boxbackup-client -> ../init.d/boxbackup-client lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc2.d/S21boxbackup-client -> ../init.d/boxbackup-client lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc2.d/S20boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc3.d/S21boxbackup-client -> ../init.d/boxbackup-client lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc3.d/S20boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc4.d/S21boxbackup-client -> ../init.d/boxbackup-client lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc4.d/S20boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc5.d/S21boxbackup-client -> ../init.d/boxbackup-client lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc5.d/S20boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 15:54 rc6.d/K21boxbackup-server -> ../init.d/boxbackup-server lrwxrwxrwx 1 root root 26 Nov 26 16:40 rc6.d/K20boxbackup-client -> ../init.d/boxbackup-client Before fixing, the two-digit numbers after the "K" and "S" were the same. This meant that the client was started before the server, since S20boxbackup-client alphanumerically precedes S20boxbackup-server. By changing the digits as I have done, the server gets started first, as it now precedes the client alphanumerically. Also, the client gets killed before the server. The links can easily be changed using the mv command. For example, #cd rc0.d #mv K20boxbackup-server K21boxbackup-server The Debian Reference manual has a pretty complete chapter on runlevels and their management. To obtain these you would need the packages debian-reference-common debian-reference debian-reference-en These are distribution independent and can be obtained from the debian web site at www.debian.org if Ubuntu does not have then in their repositories. Hope this helps. Jack Steve wrote: > I'm using the default scripts to run boxbackup client and server from > Ubuntu Server 10.10, and - as far as I can tell - the client is starting > before the server... and is failing to connect - hence generating an > error message. > > Is the client intended to delay startup until the server has started > where both are installed on the same host? If not, would it be easy to > implement it as a feature (the fewer errors in log files that need to be > manually verified, the better.) > _______________________________________________ > Boxbackup mailing list > Boxbackup at boxbackup.org > http://lists.boxbackup.org/cgi-bin/mailman/listinfo/boxbackup > -- Jack Warkentin, phone 902-404-0457, email jwark at eastlink.ca 39 Inverness Avenue, Halifax, Nova Scotia, Canada, B3P 1X6 From chris at qwirx.com Wed Mar 30 20:47:30 2011 From: chris at qwirx.com (Chris Wilson) Date: Wed, 30 Mar 2011 20:47:30 +0100 (BST) Subject: [Box Backup] Reboot server.... "unknown error (backup-error)" In-Reply-To: <4D934DC5.2020406@shic.co.uk> References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> <4D833EB3.4020002@shic.co.uk> <4D934DC5.2020406@shic.co.uk> Message-ID: Hi Steve, On Wed, 30 Mar 2011, Steve wrote: > I'm using the default scripts to run boxbackup client and server from > Ubuntu Server 10.10, and - as far as I can tell - the client is starting > before the server... and is failing to connect - hence generating an > error message. > > Is the client intended to delay startup until the server has started > where both are installed on the same host? If not, would it be easy to > implement it as a feature (the fewer errors in log files that need to be > manually verified, the better.) This is an unusual configuration. Normally the point of using an encrypted online backup system would be that the backups are stored somewhere else, not on the same system :) So why are you using Box Backup this way? Also, which "default" scripts are you using? "make install" doesn't install any initscripts, so how did they get there? Are you using someone's package? Finally, do you know *why* bbackupd starts first? Do the scripts have the same priority in /etc/rc*.d or something like that? Cheers, Chris. -- _____ __ _ \ __/ / ,__(_)_ | Chris Wilson Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer | \__/_/_/_//_/___/ | We are GNU : free your mind & your software | From sjh_boxbackup at shic.co.uk Wed Mar 30 23:16:22 2011 From: sjh_boxbackup at shic.co.uk (Steve) Date: Wed, 30 Mar 2011 23:16:22 +0100 Subject: [Box Backup] Reboot server.... "unknown error (backup-error)" In-Reply-To: References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> <4D833EB3.4020002@shic.co.uk> <4D934DC5.2020406@shic.co.uk> Message-ID: <4D93ABB6.1090306@shic.co.uk> On 30/03/11 20:47, Chris Wilson wrote: > This is an unusual configuration. Normally the point of using an > encrypted online backup system would be that the backups are stored > somewhere else, not on the same system :) So why are you using Box > Backup this way? Ideally I'd have backups on a different unix/linux/bsd box, but I only have one always-on system meeting this criteria available to my home-office LAN. I still get a lot of benefits... in spite of a lack of physical separation. I can still write the backups to external drives; I don't need to panic if the backup gets stolen; I can recover even if I accidentally delete files, or files get overwritten by some rogue process. I agree - it would be even better if I had two physical locations with physically distinct servers - but just using one server is better than no continuous backup solution at all... and, perhaps, in future, I'll have more kit at my disposal. > Also, which "default" scripts are you using? "make install" doesn't > install any initscripts, so how did they get there? Are you using > someone's package? Yes, I'm using the standard package for Ubuntu. It gives me: /etc/init.d/boxbackup-server and /etc/init.d/boxbackup-client which form part of the (still supported) legacy start-up scripts. It is generally recommended for new services to use the new 'Upstart' infrastructure which, as an advantage over the elder rc.d approach, supports arbitrary 'events' to better model service dependencies. > Finally, do you know *why* bbackupd starts first? Do the scripts have > the same priority in /etc/rc*.d or something like that? I imagine so... I've not investigated in any further detail yet... From sjh_boxbackup at shic.co.uk Wed Mar 30 23:22:56 2011 From: sjh_boxbackup at shic.co.uk (Steve) Date: Wed, 30 Mar 2011 23:22:56 +0100 Subject: [Box Backup] Reboot server.... "unknown error (backup-error)" In-Reply-To: <4D9376C7.5040901@eastlink.ca> References: <4D80E2B8.8080908@shic.co.uk> <4D81360F.8070807@shic.co.uk> <4D833EB3.4020002@shic.co.uk> <4D934DC5.2020406@shic.co.uk> <4D9376C7.5040901@eastlink.ca> Message-ID: <4D93AD40.4030705@shic.co.uk> On 30/03/11 19:30, Jack Warkentin wrote: > Before fixing, the two-digit numbers after the "K" and "S" were the > same. This meant that the client was started before the server, since > S20boxbackup-client alphanumerically precedes S20boxbackup-server. By > changing the digits as I have done, the server gets started first, as > it now precedes the client alphanumerically. Also, the client gets > killed before the server. Hi Jack, Yes, I can confirm that the two-digit numbers after K and S are all '20' on my system. This fix looks like something that would be best rolled into the package - so that all users get the advantage, and we existing users don't have to worry about patches when we upgrade... Do we know who maintains the package? Steve