Return-Path: X-Original-To: apmail-subversion-users-archive@minotaur.apache.org Delivered-To: apmail-subversion-users-archive@minotaur.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id EAE9E1070D for ; Mon, 10 Mar 2014 06:25:48 +0000 (UTC) Received: (qmail 69994 invoked by uid 500); 10 Mar 2014 06:25:48 -0000 Delivered-To: apmail-subversion-users-archive@subversion.apache.org Received: (qmail 69602 invoked by uid 500); 10 Mar 2014 06:25:42 -0000 Mailing-List: contact users-help@subversion.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Delivered-To: mailing list users@subversion.apache.org Delivered-To: moderator for users@subversion.apache.org Received: (qmail 60033 invoked by uid 99); 9 Mar 2014 17:52:16 -0000 X-ASF-Spam-Status: No, hits=0.0 required=5.0 tests=RCVD_IN_DNSWL_NONE,SPF_PASS,T_FILL_THIS_FORM_SHORT X-Spam-Check-By: apache.org Received-SPF: pass (athena.apache.org: local policy includes SPF record at spf.trusted-forwarder.org) X-TCPREMOTEIP: 71.220.4.134 X-Authenticated-UID: sstadler@incircuits.com Reply-To: From: "Steven Stadler" To: Subject: Corrupt Node - revision '0.0.r528/72560508' Date: Sun, 9 Mar 2014 12:51:39 -0500 Organization: InCircuits, Inc. Message-ID: <002d01cf3bc0$39d53f30$ad7fbd90$@incircuits.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Outlook 15.0 Thread-Index: Ac87v/JUHiMSASRVS86N2Kcrw5hstw== Content-Language: en-us X-Virus-Checked: Checked by ClamAV on apache.org The person from the CollabNet forum suggested posting this here, " It might make sense to post this on the users at subversion dot apache dot org list. I have seen developers offer to look at corrupt revision files and fix them by hand if possible." Thank You, Steve Corrupt Node - revision '0.0.r528/72560508' Author sstadler (OCN Member) Full name Steven Stadler Date 2014-03-09 08:42:06 PDT Message I had a hard drive failure, where the drive reported that the data should not be corrupt. I ordered new drives, and while putting my data on to the new drives I found that two of my SVN db files were corrupt, and my automatic backup had run so the backup was corrupt as well. I run verify, and the report indicates that revision 528 is corrupt. I have a backup dump for 0:597. I tried to make a dump for 598:665 so I could use those two dumps to make a new repository, but it does not do anything? Maybe there is also a corrupt revision within 598:665? How do I verify a range following a corrupt revision? If I load the backup dump for 0:597, will it overwrite revisions 0:597, and keep the rest 598:665? Can I just load or recover revision 528 from a dump? Thank You, Author markphip (OCN Member) Full name Mark Phippard Date 2014-03-09 09:02:35 PDT Message On Mar 9, 2014, at 11:42 AM, Steven Stadler wrote: > I had a hard drive failure, where the drive reported that the data should not be corrupt. I ordered new drives, and while putting my data on to the new drives I found that two of my SVN db files were corrupt, and my automatic backup had run so the backup was corrupt as well. > > I run verify, and the report indicates that revision 528 is corrupt. It might make sense to post this on the users at subversion dot apache dot org list. I have seen developers offer to look at corrupt revision files and fix them by hand if possible. > > I have a backup dump for 0:597. > > I tried to make a dump for 598:665 so I could use those two dumps to make a new repository, but it does not do anything? What do you mean by that? Did you add the --incremental option when you ran this? > Maybe there is also a corrupt revision within 598:665? How do I verify a range following a corrupt revision? Did you try this? $ svnadmin verify -r 598:HEAD /path/to/repos > If I load the backup dump for 0:597, will it overwrite revisions 0:597, and keep the rest 598:665? No. You could only load that dump file into a new, empty repository. So you would need that second dump file, made with the --incremental option, so that you could load the rest of the revisions. > Can I just load or recover revision 528 from a dump? If it came to that, you would probably need to load the entire repos from the dump file. You could try creating a new empty repository and load your dump file into it. Then compare the revision files for r528. It might be possible to just copy those files into the old repository. It will depend a lot on using the same version of SVN though. One of the reasons people dump/load is that the on-disk format "improves" with new releases by using newer techniques for storing the data. So if you are loading the repository with a newer version of SVN than created the original repository, it is unlikely you could simply "drop" a revision file from the new version to the old one. That technique is not supported anyway, I am just saying it might be possible and might work. The revision files are stored in the db/revs and db/revprops folders. So look for: db/revs/0/528 and db/revprops/0/528 If you have a really old repository format, then the /0/ folder will not be there. Mark Author sstadler (OCN Member) Full name Steven Stadler Date 2014-03-09 09:55:42 PDT Message Thanks Mark!! I tried making a new repository from the dump 0:597, and I copied the files from the db\revs and db\revprops for 528 from the new repository to the old one, where the 528 file in the old db\revs had a very different file size than the new one. The repository still does not verify and it says that revision 527 is now corrupt, it has a malformed header. I was thinking about copying over 527, and see if 526 will then be bad?? but 527 was good before copying over 528? I should post this on subversion dot apache dot org list before I make things worse.... I also tried making a dump of 598:665 selecting the incremental check box, and it still did not make a dump?? I also tried svnadmin verify -r 598:HEAD /path/to/repos and it returns Can't read length in line in file /path/to/repos \db\revs\528 Is there a way to forward this thread to the users at subversion dot apache dot org list? Thanks again, Steve Author sstadler (OCN Member) Full name Steven Stadler Date 2014-03-09 10:04:30 PDT Message just another not, I just tried to make a dump of 598:665 from the CMD window instead if from the CollabNet web interface, and the dump returns the same error as the verify, Can't read length in line in file /path/to/repos \db\revs\528 Steven R Stadler PE InCircuits Inc 4284 Reiland Lane Shoreview MN 55126 Tel 651 765 4225 Fax 651 765 4226 Cell 651 492 9792 sstadler@incircuits.com www.incircuits.com