Back to bug 2207491
| Who | When | What | Removed | Added |
|---|---|---|---|---|
| Red Hat One Jira (issues.redhat.com) | 2023-05-16 06:22:30 UTC | Link ID | Red Hat Issue Tracker RHCEPH-6668 | |
| Xiubo Li | 2023-05-16 06:22:56 UTC | Target Release | 6.1z1 | 5.3z4 |
| Assignee | vshankar | xiubli | ||
| Xiubo Li | 2023-05-16 06:23:13 UTC | Status | NEW | ASSIGNED |
| Xiubo Li | 2023-05-16 06:24:23 UTC | Link ID | Ceph Project Bug Tracker 52280 | |
| Veera Raghava Reddy | 2023-05-23 03:13:20 UTC | Severity | unspecified | high |
| CC | vereddy | |||
| Greg Farnum | 2023-05-23 13:48:16 UTC | Flags | needinfo?(xiubli) | |
| CC | gfarnum | |||
| Xiubo Li | 2023-05-23 13:53:07 UTC | Flags | needinfo?(xiubli) | |
| Hemanth Kumar | 2023-05-30 05:36:24 UTC | QA Contact | hyelloji | amk |
| Greg Farnum | 2023-06-13 06:53:12 UTC | CC | bkunal | |
| Greg Farnum | 2023-06-14 14:50:45 UTC | Status | ASSIGNED | POST |
| Greg Farnum | 2023-06-14 16:10:49 UTC | Flags | needinfo?(xiubli) | |
| Flags | needinfo?(gfarnum) | |||
| CC | tserlin | |||
| Flags | needinfo?(gfarnum) | needinfo?(tserlin) | ||
| Greg Farnum | 2023-06-14 21:58:07 UTC | Flags | needinfo?(tserlin) | |
| Flags | needinfo?(xiubli) needinfo?(mchangir) needinfo?(vshankar) | |||
| CC | mchangir, vshankar | |||
| Xiubo Li | 2023-06-15 01:31:41 UTC | Flags | needinfo?(xiubli) needinfo?(xiubli) needinfo?(mchangir) needinfo?(vshankar) | |
| errata-xmlrpc | 2023-06-15 15:56:54 UTC | Status | POST | MODIFIED |
| Fixed In Version | ceph-16.2.10-183.el8cp | |||
| Status | MODIFIED | ON_QA | ||
| Akash Raj | 2023-06-16 08:55:14 UTC | CC | akraj | |
| Blocks | 2210690 | |||
| Docs Contact | akraj | |||
| Flags | needinfo?(xiubli) | |||
| Xiubo Li | 2023-06-16 23:13:42 UTC | Doc Text | Cause: When replaying the journals, if the inodetable or the sessionmap versions do not match with the ones in the MDS caches, the corresponding CInodes will be added to the inode_map, but without removing the ino# from the inodetable or sessions' prealloc inos list. Consequence: Will cause allocating CInode fail because it has already in the inode map. Fix: So when allocating new CInode we should skip the corresponding ino# even replaying journal fails. Result: Allcating inodes will succeed. | |
| Doc Type | If docs needed, set a value | Bug Fix | ||
| Xiubo Li | 2023-06-16 23:16:45 UTC | Flags | needinfo?(xiubli) | |
| Amarnath | 2023-06-26 19:17:18 UTC | Flags | needinfo?(xiubli) | |
| Xiubo Li | 2023-06-27 00:28:13 UTC | Flags | needinfo?(xiubli) | needinfo?(amk) |
| CC | amk | |||
| Akash Raj | 2023-06-27 04:16:27 UTC | Doc Text | Cause: When replaying the journals, if the inodetable or the sessionmap versions do not match with the ones in the MDS caches, the corresponding CInodes will be added to the inode_map, but without removing the ino# from the inodetable or sessions' prealloc inos list. Consequence: Will cause allocating CInode fail because it has already in the inode map. Fix: So when allocating new CInode we should skip the corresponding ino# even replaying journal fails. Result: Allcating inodes will succeed. | .Allocating inodes no longer fails Previously, when replaying the journals, if the `inodetable` or the `sessionmap` versions did not match with the ones in the MDS caches, the corresponding `CInodes` would be added to the _inode_map_, but without removing the `ino#` from the `inodetable` or the `sessions`'s prealloc inos list. This would cause the allocation of `CInode` to fail as it was already added in the inode map. With this fix, when allocating new `CInode`, the corresponding `ino#` is skipped, even when replaying the journal fails and allocating inodes succeeds. |
| Flags | needinfo?(xiubli) | |||
| Xiubo Li | 2023-06-27 04:35:23 UTC | Flags | needinfo?(xiubli) | |
| Amarnath | 2023-06-27 06:06:55 UTC | Flags | needinfo?(amk) | |
| Amarnath | 2023-06-27 06:07:17 UTC | Flags | needinfo?(xiubli) | |
| Xiubo Li | 2023-06-27 06:22:56 UTC | Flags | needinfo?(xiubli) | |
| Amarnath | 2023-06-27 08:47:45 UTC | Flags | needinfo? | |
| Status | ON_QA | VERIFIED | ||
| Akash Raj | 2023-06-28 06:57:07 UTC | Doc Text | .Allocating inodes no longer fails Previously, when replaying the journals, if the `inodetable` or the `sessionmap` versions did not match with the ones in the MDS caches, the corresponding `CInodes` would be added to the _inode_map_, but without removing the `ino#` from the `inodetable` or the `sessions`'s prealloc inos list. This would cause the allocation of `CInode` to fail as it was already added in the inode map. With this fix, when allocating new `CInode`, the corresponding `ino#` is skipped, even when replaying the journal fails and allocating inodes succeeds. | .Allocating inodes no longer fails Previously, when replaying the journals, if the `inodetable` or the `sessionmap` versions did not match with the ones in the MDS caches, the corresponding `CInodes` would be added to the _inode_map_, but without removing the `ino#` from the `inodetable` or the `sessions`'s prealloc inos list. This would cause the allocation of `CInode` to fail as it was already added in the inode map. With this fix, when allocating new `CInode`, the corresponding `ino#` is skipped. |
| Akash Raj | 2023-07-18 05:40:08 UTC | Doc Text | .Allocating inodes no longer fails Previously, when replaying the journals, if the `inodetable` or the `sessionmap` versions did not match with the ones in the MDS caches, the corresponding `CInodes` would be added to the _inode_map_, but without removing the `ino#` from the `inodetable` or the `sessions`'s prealloc inos list. This would cause the allocation of `CInode` to fail as it was already added in the inode map. With this fix, when allocating new `CInode`, the corresponding `ino#` is skipped. | .Allocating inodes no longer fails Previously, when replaying the journals, if the `inodetable` or the `sessionmap` versions did not match with the ones in the MDS caches, the corresponding `CInodes` would be added to the _inode_map_, but without removing the `ino#` from the `inodetable` or the `sessions` prealloc inos list. This would cause the allocation of `CInode` to fail as it was already added in the inode map. With this fix, when allocating new `CInode`, the corresponding `ino#` is skipped. |
| errata-xmlrpc | 2023-07-19 15:42:03 UTC | Status | VERIFIED | RELEASE_PENDING |
| errata-xmlrpc | 2023-07-19 16:19:11 UTC | Status | RELEASE_PENDING | CLOSED |
| Resolution | --- | ERRATA | ||
| Last Closed | 2023-07-19 16:19:11 UTC | |||
| errata-xmlrpc | 2023-07-19 16:20:03 UTC | Link ID | Red Hat Product Errata RHBA-2023:4213 |
Back to bug 2207491