| Summary: | UFS write becomes very slow in some cases | ||
|---|---|---|---|
| Product: | Base System | Reporter: | Dmitry Sivachenko <demon> |
| Component: | kern | Assignee: | freebsd-fs (Nobody) <fs> |
| Status: | New --- | ||
| Severity: | Affects Many People | CC: | mckusick |
| Priority: | --- | ||
| Version: | 10.1-STABLE | ||
| Hardware: | Any | ||
| OS: | Any | ||
Hello, I have 94TB h/w raid6 volume with UFS filesystem on it: tunefs: POSIX.1e ACLs: (-a) disabled tunefs: NFSv4 ACLs: (-N) disabled tunefs: MAC multilabel: (-l) disabled tunefs: soft updates: (-n) enabled tunefs: soft update journaling: (-j) disabled tunefs: gjournal: (-J) disabled tunefs: trim: (-t) disabled tunefs: maximum blocks per file in a cylinder group: (-e) 4096 tunefs: average file size: (-f) 16384 tunefs: average number of files in a directory: (-s) 64 tunefs: minimum percentage of free space: (-m) 8% tunefs: space to hold for metadata blocks: (-k) 9136 tunefs: optimization preference: (-o) time tunefs: volume label: (-L) When it is about 5TB free on it (which does not count 8% of reserved space) /dev/mfid0p1 94T 82T 5.1T 94% /opt all write operations on it become very-very slow (1-2 MB/sec). The process which writes data consumes 100% of CPU in system state: {agata}/home/mitya% ps -p 58796 -o systime,usertime SYSTIME USERTIME 0:59.56 0:00.08 58796 mitya 1 103 0 27808K 4820K CPU11 11 0:53 100.00% bsdtar As soon as I free some space, the problem disappears until space fills up again. Since I have the recommended 8% of reserved free space, I expect writes to work well until all free space is exhausted.