On 04/25/2012 12:28 PM, Christian Brunner wrote: > I've tried test_filestore_workloadgen, but it didn't work very well. > After 5 Minutes it terminated with the following messages: > > 2012-04-25 11:28:57.768709 7fcac3f69760 0 Destroying collection > '0.1_head' (358 objects) > 2012-04-25 11:29:07.478747 7fcac3f69760 0 Destroying collection > '0.22_head' (477 objects) > 2012-04-25 11:29:07.479035 7fcac3f69760 0 We ran out of collections! > 2012-04-25 11:29:07.916149 7fcac3f69760 1 journal close > /dev/mapper/vg01-lv_osd_journal_0 > > I guess this is not btrfs related .-) > Oh yeah... the "solution" for that is to set '--test-num-colls' to a greater value (the test creates 30 by default, and destroys them every 'X' transactions). Increase the number of collections and it will have a longer run. But since that's sounding like a workaround over something that could easily be avoided, I'll make a patch for the test to avoid that from happening. -- João Eduardo Luís gpg key: 477C26E5 from pool.keyserver.eu
Attachment:
signature.asc
Description: OpenPGP digital signature
