View Issue Details

IDProjectCategoryView StatusLast Update
0003165GNUnetnamestore servicepublic2013-12-24 19:53
ReporterChristian Grothoff Assigned ToLRN Due Date
PriorityhighSeveritymajorReproducibilityhave not tried
Status closedResolutionfixed 
Product VersionGit master 
Target Version0.10.0Fixed in Version0.10.0 
Summary0003165: namestore tests fail on W32
Descriptionhttps://gnunet.org/buildbot/builders/polot-win7-i386/builds/1058/steps/tests%20namestore/logs/stdio
TagsNo tags attached.

Activities

Christian Grothoff

2013-12-08 21:39

manager   ~0007789

Last edited: 2013-12-08 21:40

22:37 <grothoffhome> LRN: are you aware of 0003165?
22:37 <LRN> yes
22:37 <LRN> i am baffled by it
22:38 <grothoffhome> :-(.
22:38 <LRN> Let's see if my recent changes fix anything
22:38 <LRN> because this test passes on my machine
22:38 <grothoffhome> I was baffled by the fact that this one test would fail for W32 and pass elsewhere, as I don't see how namestore should be W32-specific...
22:39 <grothoffhome> But good to know that it is passing on your system, so maybe it is something on the VM itself.
22:39 <LRN> until 5 minutes ago my system had +5 patches on top of SVN HEAD
22:39 <LRN> This might have been the reason :)
22:40 <grothoffhome> Well, the next run of the BB will tell...

LRN

2013-12-09 02:20

reporter   ~0007793

Last edited: 2013-12-09 02:20

Found the problem.
Tests passed on my machine, because i have somewhat more extensive pre-test cleanup procedures than buildslave does.
In particular, these procedures remove namestore tmp directory, the one with test_sqlite.* database files.
These files are supposed to be removed by the testcase, but it fails to do so, because files are in use by the namestore service. For some reason the testcase tries to clean them up (by using GNUNET_DISK_directory_remove()) before arm & namestore are dead.
Subsequent tests fail because database from previous test leaks into them.

LRN

2013-12-09 14:41

reporter   ~0007798

Moving cleanup to the end of main() alleviates the problem for me (only tried this on one test though).

LRN

2013-12-11 16:55

reporter   ~0007843

Fixed in r31231

Issue History

Date ModifiedUsernameFieldChange
2013-12-08 19:00Christian GrothoffNew Issue
2013-12-08 19:00Christian GrothoffStatusnew => assigned
2013-12-08 19:00Christian GrothoffAssigned To => Matthias Wachs
2013-12-08 21:39Christian GrothoffNote Added: 0007789
2013-12-08 21:40Christian GrothoffNote Edited: 0007789
2013-12-09 02:20LRNNote Added: 0007793
2013-12-09 02:20LRNNote Edited: 0007793
2013-12-09 14:41LRNNote Added: 0007798
2013-12-11 16:55LRNNote Added: 0007843
2013-12-11 16:56LRNStatusassigned => resolved
2013-12-11 16:56LRNFixed in Version => Git master
2013-12-11 16:56LRNResolutionopen => fixed
2013-12-11 16:56LRNAssigned ToMatthias Wachs => LRN
2013-12-11 19:10Christian GrothoffFixed in VersionGit master => 0.10.0
2013-12-24 19:53Christian GrothoffStatusresolved => closed