bzr: ERROR: Error in data for index GraphIndex
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Bazaar |
Expired
|
Undecided
|
Unassigned |
Bug Description
I get the following error when trying to continue pulling from an SVN repository.
$ bzr pull -r 6700
Using saved parent location: https:/
bzr: ERROR: Error in data for index GraphIndex(
$
And had the following conversation in #bzr:
(10:06:18 PM) Ted: lifeless: Hey, is there a way to regenerate the indices?
(10:06:32 PM) Ted: I'm getting an odd error: bzr: ERROR: Error in data for index GraphIndex(
(10:12:23 PM) spiv: tedg: I haven't seen that error before...
(10:12:57 PM) ***Ted seems to have special skills in breaking bazaar :(
(10:22:25 PM) spiv: tedg: off the top of my head I don't think it's possible to regenerate a .tix. I certainly don't know of any convenient tools to do it.
(10:22:58 PM) spiv: tedg: which repo format? pack-0.92?
spiv spm
(10:23:25 PM) Ted: spiv: What ever is default in the nightlies plus rich-root-pack.
(10:23:40 PM) ***spiv nods
(10:23:59 PM) lifeless: tedg: file a bug; we'll need to write a repair tool to regen the index for you.
(10:24:07 PM) lifeless: tedg: the index is zlib'd data
(10:24:39 PM) Ted: lifeless: What should be in the bug, just that single tix file or the whole repo?
(10:24:43 PM) lifeless: tedg: its mostly (but not entirely) derivable from what you have on disk
(10:25:00 PM) lifeless: tedg: the one tix which may let us repair it depending on the damage.
(10:25:09 PM) lifeless: tedg: and this conversation probably
(10:25:12 PM) Ted: I mean, I'm just trying to get a checkout from SF SVN -- but that takes days and this happened in the middle.
(10:25:27 PM) Ted: I can start the process over, I don't need that tix specifically.
(10:25:42 PM) Ted: I'm just trying to avoid starting over.
(10:25:43 PM) lifeless: the way we could repair is to read the leaf pages only to get back the kys and values and then reoutput it.
(10:26:09 PM) lifeless: tedg: if this occured during a fresh branch from svn, well check your hardware - you may have a disk or memory problem
(10:26:49 PM) spiv: lifeless: a little bit odd to get BadIndexError rather than a zlib error if it's bad hardware, though.
(10:27:13 PM) lifeless: spiv: depends on the try's.
(10:27:21 PM) lifeless: also we can look at the index once the bug is filed.
(10:28:22 PM) Ted: So, I guess I'm not 100% clear what I should attach to the bug.
(10:28:31 PM) lifeless: 13:19 < lifeless> tedg: the one tix which may let us repair it depending on the damage.
(10:28:34 PM) lifeless: 13:19 < lifeless> tedg: and this conversation probably
(10:28:54 PM) Ted: lifeless: Okay.
And you might like this information as well:
$ apt-cache policy bzr bzr-svn
bzr:
Installed: 1.15+4431+110
Candidate: 1.15+4431+110
Version table:
*** 1.15+4431+110 0
500 http://
100 /var/lib/
1.13.1-1 0
500 http://
bzr-svn:
Installed: 0.6~3053~330~09.04
Candidate: 0.6~3053~330~09.04
Version table:
*** 0.6~3053~330~09.04 0
500 http://
100 /var/lib/
0.5.3-1 0
500 http://
(I guess it's a rich-root-pack format repo, as the indices aren't in btree (1.9) format.)
Line 451 of that .tix file is pretty obviously corrupted; it's filled with about 5313 bytes of binary garbage of some sort. It seems unlikely that bzr would have written that, so my wild guess would be that there's a hardware or filesystem issue.