On Sun, Dec 8, 2019 at 10:38 PM Daniel Shahaf <d.s_at_daniel.shahaf.name> wrote:
>
> Nathan Hartman wrote on Sun, 08 Dec 2019 06:04 +00:00:
> > Commit can fail with an undeserved "Filesystem is corrupt" error
> >
> > See issue 4722, "checksum fail during commit when delta is 16K".
> >
> > Under a very specific set of circumstances, a commit may fail with an
> > undeserved "Filesystem is corrupt" error. This bug affects the 1.9.7
> > release only. A fix for this problem has been included in the 1.9.9
> > release.
>
> Anyone reading this is going to wonder if either version number is a
> typo. Clarify that they aren't? E.g., say that "1.9.8 was not
> publicly released".
Agreed. More below:
> Having the error's number and the specific error message, if any, will be
> helpful. If they are stated on the JIRA issue or on the thread, it would be
> nice to quote them in the release notes. If they're not immediately available,
> however, in principle we could reconstruct them by reproducing the bug with
> 1.9.7; however, in light of 1.9's EOL schedule, I wouldn't consider it high
> priority to reconstruct those details.
>
> The string printed is svn_strerror(SVN_ERR_FS_CORRUPT), so the error number is
> probably E160004.
The original email to users@ mentions the error:
[[[
D:\confidential>svn commit secret.bin -m "Testing broken commit"
Sending secret.bin
Transmitting file data .svn: E160000: Commit failed (details follow):
svn: E160000: SHA1 of reps '604440 34 134255 136680
c9f4fabc4d093612fece03c339401058
db11617ef1454332336e00abc311d44bc698f3b3 605556-czmh/_8' and '-1 0
134255 136680 c9f4fabc4d093612fece03c339401058
db11617ef1454332336e00abc311d44bc698f3b3 605556-czmh/_8' matches
(db11617ef1454332336e00abc311d44bc698f3b3) but contents differ
]]]
Link:
https://svn.haxx.se/users/archive-2018-02/0098.shtml
also at:
http://mail-archives.apache.org/mod_mbox/subversion-users/201802.mbox/%3cCAJM5Lfu2+fDDmmc4938LrsF=UHEuRY9cprf2LKvg0CsCVp36AQ@mail.gmail.com%3e
I'm not sure where the "filesystem is corrupt" idea comes from. It may
have just been caused by people describing the problem in loose terms.
So I think I'll write:
[[[
Index: 1.9.html
===================================================================
--- 1.9.html (revision 1871084)
+++ 1.9.html (working copy)
@@ -1466,6 +1466,26 @@
</div> <!-- shattered-sha1 -->
+<div class="h3" id="svn-4722">
+<h3>Commit can fail with an undeserved "Filesystem is corrupt" error
+ <a class="sectionlink" href="#svn-4722"
+ title="Link to this section">¶</a>
+</h3>
+
+<p>See issue 4722, "checksum fail during commit when delta is 16K".
+</p>
+
+<p>When using a Subversion 1.9.7 server, a commit may fail with an
+undeserved SHA1 collision error: "E160000: SHA1 of reps ... and ...
+matches (...) but contents differ." For details, see
+<a href="https://issues.apache.org/jira/browse/SVN-4722?issueNumber=4722"
+>issue 4722, "checksum fail during commit when delta is 16K"</a>. This
+bug affects the 1.9.7 release only. A fix for this problem has been
+included in the 1.9.9 release (1.9.8 was not publicly released).
+</p>
+
+</div> <!-- svn-4722 -->
+
</div> <!-- issues -->
<div class="h2" id="troubleshooting">
]]]
I know it's not terribly high priority to go back and fill in old
release notes but it seems to me (based on discussions and questions
we get) that many people are still using 1.9.7 on the server. Some are
using older OS distros. Others are using Subversion packages that
install onto a NAS. In any case, they have 1.9.7 because they haven't
received updates. Maybe they don't see any issues in the release notes
affecting versions later than 1.9.6 so they don't think it's important
to update.
There are other bugs fixes in later 1.9.x as well. It's better to
write it down than to keep it in my head.
Nathan
Received on 2019-12-09 16:43:18 CET