Description of problem: After migrating data from a distributed-stripe using remove-brick mount point shows different checksum across the arequal runs Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. Create a distributed-stripe volume 3x2 2. add some data on the mount point and calculate arequal cheksum 3. Remove brick start on any one of the pair , once remove-brick completes commit it. 4. Now again calculate areequal checksum Actual results: Checksum differs. Expected results: Additional info: Client logs says [2012-05-21 19:52:22.155933] E [stripe-helpers.c:271:stripe_ctx_handle] 3-dis-stripe-stripe-2: Failed to get stripe-size [2012-05-21 19:52:22.155962] E [stripe.c:212:stripe_lookup_cbk] 3-dis-stripe-stripe-2: Error getting fctx info from dict [2012-05-21 19:52:22.156118] E [stripe-helpers.c:271:stripe_ctx_handle] 3-dis-stripe-stripe-2: Failed to get stripe-size [2012-05-21 19:52:22.156141] E [stripe.c:212:stripe_lookup_cbk] 3-dis-stripe-stripe-2: Error getting fctx info from dict [2012-05-21 19:52:22.158532] W [page.c:984:__ioc_page_error] 3-dis-stripe-io-cache: page error for page = 0x7f3ec8038fa0 & waitq = 0x7f3ec8034b70 [2012-05-21 19:52:22.158654] W [page.c:984:__ioc_page_error] 3-dis-stripe-io-cache: page error for page = 0x7f3ec8006f20 & waitq = 0x7f3ec801ac30 [2012-05-21 19:52:22.159257] E [stripe-helpers.c:271:stripe_ctx_handle] 3-dis-stripe-stripe-2: Failed to get stripe-size [2012-05-21 19:52:22.159297] E [stripe.c:212:stripe_lookup_cbk] 3-dis-stripe-stripe-2: Error getting fctx info from dict [2012-05-21 19:52:22.159508] E [stripe-helpers.c:271:stripe_ctx_handle] 3-dis-stripe-stripe-2: Failed to get stripe-size [2012-05-21 19:52:22.159530] E [stripe.c:212:stripe_lookup_cbk] 3-dis-stripe-stripe-2: Error getting fctx info from dict [2012-05-21 19:52:22.162160] W [page.c:984:__ioc_page_error] 3-dis-stripe-io-cache: page error for page = 0x7f3ec8038fa0 & waitq = 0x7f3ec8034b70
Can we check if this issue still exists?
This issue is not reproducible on 3.4.0qa5-1.el6rhs.x86_64
Because of posix_mknod is not sending the xattr filled while unwinding, it couldn't get the stripe-size from the xattr. So, I think, mostly the patch http://review.gluster.org/3904 would fix the issue.
pre-release version is ambiguous and about to be removed as a choice. If you believe this is still a bug, please change the status back to NEW and choose the appropriate, applicable version for it.