Fix for zfs > 2.2.0 (closes #135) #136
Conversation
- more reliable and readable stats file parsing - uses original logic for older versions of zfs - I'm not convinced the original logic is right after reading https://utcc.utoronto.ca/~cks/space/blog/solaris/ZFSARCItsVariousSizes, but I don't want to potentially break things
|
@flaviut Could you add a description please? What has changed in ZFS and how was the problem solved in this PR? Will compatibility with older versions of ZFS be violated after merging this PR? |
|
There's more explanation about what's changed at openzfs/zfs@a8d83e2, although that goes into too much detail and is hard to understand. In short, new zfs no longer keeps metadata in a separate pool in the ARC. For old versions of zfs, exact same old behavior is retained. That commit message implies that there is almost no un-evictable metadata in the ARC in new zfs, so we don't need to consider it anymore (and we don't actually have the data to consider it if we wanted to). with regards to |
| # return c_min, size, arc_meta_used, arc_meta_min, zfs_available | ||
| # old zfs: consider arc_meta_used, arc_meta_min | ||
| if 'arc_meta_used' in stats and 'arc_meta_min' in stats: | ||
| meta_rec = max(stats['arc_meta_used'] - stats['arc_meta_min'], 0) |
There was a problem hiding this comment.
this line is semantically exactly the same as the previous version:
meta_rec = arc_meta_used - arc_meta_min
if meta_rec < 0:
meta_rec = 0
| try: | ||
| # find indexes | ||
| with open(arcstats_path, 'rb') as f: | ||
| a_list = f.read().decode().split('\n') | ||
| for n, line in enumerate(a_list): | ||
| if line.startswith('c_min '): | ||
| c_min_index = n | ||
|
|
||
| elif line.startswith('size '): | ||
| size_index = n | ||
|
|
||
| elif line.startswith('arc_meta_used '): | ||
| arc_meta_used_index = n | ||
|
|
||
| elif line.startswith('arc_meta_min '): | ||
| arc_meta_min_index = n | ||
|
|
||
| else: | ||
| continue | ||
| except Exception as e: | ||
| log(e) | ||
| ZFS = False |
There was a problem hiding this comment.
instead of storing these into individual variables and writing fairly complicated code to parse this file again, we have a method that stores the whole contents of the file into an easy-to-read dict.
|
Merged, thanks! |
|
thank you for your work writing and maintaining this project! |
No description provided.