[DOCS-7104] EXT4 and WT incompatibilities Created: 05/Feb/16 Updated: 30/Oct/23 Resolved: 18/Feb/20 |
|
| Status: | Closed |
| Project: | Documentation |
| Component/s: | Server |
| Affects Version/s: | None |
| Fix Version/s: | Server_Docs_20231030 |
| Type: | Task | Priority: | Critical - P2 |
| Reporter: | Osmar Olivo | Assignee: | Brian Lane |
| Resolution: | Done | Votes: | 4 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||
| Participants: | |||||||||||||
| Days since reply: | 3 years, 48 weeks, 5 days ago | ||||||||||||
| Story Points: | 0 | ||||||||||||
| Description |
|
Currently we are lacking in a comprehensive explanation of the issues that can arise with WT when running on EXT4. Specifically, details on what the issue is, under what conditions it can manifest, and what the symptoms are. We have similar pages explaining how NUMA or low ulimits can negatively impact a cluster, and therefore it makes sense to have a page covering EXT4's issue in a bit more detail. Particularly since it is a widely accepted filesystem and the issues will only manifest themselves under certain conditions. |
| Comments |
| Comment by Nic Cottrell [ 06/Mar/20 ] |
|
brian.lane - did you mean to close this DOCS ticket? If I understand the SERVER-181314 discussion correctly, it confirms that there is still EXT4 issues under heavy load. I still think a note in the docs along the line of josef.ahmad's last comment would help inform customers. Maybe something like: > Performance testing has demonstrated that when the volume of IO requests exceed the storage throughput capacity, XFS degrades the throughput more gracefully, whereas ext4 continues "full on" until it stalls altogether. For most production use cases the graceful slowdown provides better application results. |
| Comment by Brian Lane [ 09/Oct/19 ] |
|
Just dropping on a note on this issue in case people have not been watching Some recent tests with RHEL 7.6 and MongoDB 4.0.12 and stalling still with EXT4. -Brian |
| Comment by Joel Parke [X] [ 25/Apr/19 ] |
|
It seems that this warning is so old, when the kernel for Ubuntu 19.04 is now 5.0, that forcing one to use XFS with WT, seems a bit lame. Especially when the OLD stalls never did express exactly why or if later kernels had this issue. Is this simply because in production, NO ONE is running XFS with WT with more modern kernels? If so this is doing a huge disservice to mongodb. If anyone has clear insight or experience with this issue, please comment!!! I never see this issue in my production system! but I wouldn't expect to, since my files are small. |
| Comment by Aaron Queen [ 31/Aug/17 ] |
|
Has there been any movement on this? Most distros still ship with EXT4 as default, and it seems strange to strongly suggest using a specific file system for your data directory without the information needed to justify going through the hassle of partitioning off with XFS. EXT4 and XFS both receive updates with new kernels as well, so is this problem specific to certain kernels under a specific version, or is it EXT4 in general, even on the very latest mainline kernel? This type of information would be very helpful, and could also possibly eliminate the startup warning if it is actually specifically EXT4 + kernel below a certain 2.x, 3.x, 4.x version. |