Bug 1752739
| Summary: | fuse mount crash observed with sharding + truncate | ||
|---|---|---|---|
| Product: | [Red Hat Storage] Red Hat Gluster Storage | Reporter: | bipin <bshetty> |
| Component: | sharding | Assignee: | Vinayak Hariharmath <vharihar> |
| Status: | CLOSED ERRATA | QA Contact: | SATHEESARAN <sasundar> |
| Severity: | high | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | rhgs-3.4 | CC: | bugs, nsoffer, pprakash, puebele, ravishankar, rhs-bugs, sabose, sasundar, seamurph, sheggodu, storage-qa-internal, vharihar |
| Target Milestone: | --- | Keywords: | Reopened, Triaged, ZStream |
| Target Release: | RHGS 3.5.z Batch Update 4 | Flags: | vharihar:
needinfo+
|
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | glusterfs-6.0-50 | Doc Type: | No Doc Update |
| Doc Text: | Story Points: | --- | |
| Clone Of: | 1751722 | Environment: | |
| Last Closed: | 2021-04-29 07:20:36 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | 1751085 | ||
| Bug Blocks: | |||
|
Description
bipin
2019-09-17 06:45:58 UTC
Steps that was done to hit this issue: 1. Complete the RHHI-V deployment 2. Create 30 vm's using pool 3. After an hour or so could see the FUSE mount crash So far, this issue is hit only with 1 QE setup and not consistently, but hit twice so far with more than ~10 attempts As RHHI-V bug is not fixed with oVirt's approach to detect storage block size, this bug is no longer required urgently with RHGS 3.5.0. @Krutika, please validate my statement. If that's true, we can drop it for RHGS 3.5.0 but target the same for RHGS 3.5.z ? (In reply to SATHEESARAN from comment #6) > As RHHI-V bug is not fixed with oVirt's approach to detect storage block > size, > this bug is no longer required urgently with RHGS 3.5.0. > > @Krutika, please validate my statement. > If that's true, we can drop it for RHGS 3.5.0 but target the same for RHGS > 3.5.z ? Please read it as 'RHHI-V bug is fixed with oVirt's approach' (In reply to SATHEESARAN from comment #6) > As RHHI-V bug is not fixed with oVirt's approach to detect storage block > size, > this bug is no longer required urgently with RHGS 3.5.0. > > @Krutika, please validate my statement. > If that's true, we can drop it for RHGS 3.5.0 but target the same for RHGS > 3.5.z ? This sounds reasonable to me. Based on comment8, removing the blocker request to include this bug for RHGS 3.5.0. Also the original issue is now fixed with RHV 4.3.6 and no longer leads to crash. But the issue is genuine that there exists the problem with sharding + truncate call. And RCA is also done by Krutika. So requesting to include the fix in RHGS 3.5.z to prevent any such future usage of shard + truncate doesn't lead to any such problems. When this issue was found, it was initally complained that the fuse mount crash was seen, later through investigation it was found to be 2 seperate problems 1. Crash in FOPs because of integer overflow 2. Sharding doesn't support truncate Currently the fix addresses only 1. Excerpt from the commit message: "This patch fixes a crash in FOPs that operate on really large sharded files where number of participant shards could sometimes exceed signed int32 max." But the other issue of sharding to support truncate is not yet available, which will be tracked with a separate bug. Based on this, updating the bug summary. Tested with RHGS 3.5.4 interim build ( glusterfs-6.0-51.el8rhgs ) 1. Created a replica 3 volume and enabled sharding 2. Created 2 fuse mounts 3. Created a file of size 512B from one of the fuse mount 4. From one fuse mount truncate that file to 0 5. From another fuse mount truncate the same file to 0 again. I/O errors are noted as sharding doesn't support truncate from day1. But there are no crashes observed. Also tested this scenario in RHHI-V setup, with 10 VMs running workloads There are no issues seen. Mistakenly passed on the information to Vinayak, that this bug is fixed in RHGS 3.5.3 which led him to close this bug. Moving this bug back to ON_QA As per comment25, this bug is verified for RHGS 3.5.4 Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (glusterfs bug fix and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2021:1462 |