getTotalMemory() - skip cgroupsv1 max memory.limit_in_bytes (SYN-4433) #3198
Conversation
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## master #3198 +/- ##
==========================================
- Coverage 97.30% 97.20% -0.11%
==========================================
Files 224 224
Lines 44823 44825 +2
==========================================
- Hits 43616 43570 -46
- Misses 1207 1255 +48
Flags with carried forward coverage won't be shown. Click here to find out more.
☔ View full report in Codecov by Sentry. |
synapse/lib/platforms/linux.py
Outdated
| valu = int(f.read()) | ||
| # Skip a known value on cgroupsv1 where there has not been | ||
| # a limit set, so we will fallback to /proc/meminfo instead. | ||
| if valu != 9223372036854771712: |
There was a problem hiding this comment.
this SO thread says this value is signed 64-bit max rounded down to nearest page: https://unix.stackexchange.com/questions/420906/what-is-the-value-for-the-cgroups-limit-in-bytes-if-the-memory-is-not-restricte.
That makes more sense than this magic value. I'd recommend changing to one of these:
0x7fffffff_ffffffff ^ 0xfff or 2**63-1 ^ 0xfff to be a little more clear. Maybe link the article?
There was a problem hiding this comment.
Even better, do we have these constants already defined somewhere? INT64_MAX and PAGESIZE or something equivalent? INT64_MAX ^ (PAGESIZE - 1) would be really clear.
There was a problem hiding this comment.
PAGESIZE is actually going to be host specific, so we cannot use it as a constant - we can resolve it from os.sysconf('SC_PAGESIZE') though. We can also assume a 64 bit LONG as well. Will fix this.
vEpiphyte
changed the title
No description provided.