Details
-
Bug
-
Resolution: Fixed
-
Minor
-
Lustre 2.9.0
-
None
-
autotest review-zfs and review-ldisfs
-
3
-
9223372036854775807
Description
sanity test 101f fails with
'misses too much pages!'
From the test_log on the client, we see
== sanity test 101f: check mmap read performance ===================================================== 02:47:36 (1468403256) /usr/bin/iozone Cancel LRU locks on lustre client to flush the client cache Reset readahead stats mmap read the file with small block size checking missing pages sanity test_101f: @@@@@@ FAIL: misses too much pages!
Here is the check that is failing in the test code:
local miss=$($LCTL get_param -n llite.*.read_ahead_stats |
get_named_value 'misses' | cut -d" " -f1 | calc_total)
[ $miss -lt 3 ] || error "misses too much pages ('$miss')!"
There’s nothing interesting in the console logs.
This test does NOT fail consistently and has failed with this error message six times since 13-July-2016:
2016-07-13 - https://testing.hpdd.intel.com/test_sets/43aec3ae-48e8-11e6-8968-5254006e85c2
2016-07-13 - https://testing.hpdd.intel.com/test_sets/af1ac4e8-4948-11e6-9f8e-5254006e85c2
2016-07-14 - https://testing.hpdd.intel.com/test_sets/c663d7fa-49b3-11e6-9f8e-5254006e85c2
2016-07-14 - https://testing.hpdd.intel.com/test_sets/95d24332-4a27-11e6-bf87-5254006e85c2
2016-07-17 - https://testing.hpdd.intel.com/test_sets/c3840482-4c02-11e6-bf87-5254006e85c2
2016-07-17 - https://testing.hpdd.intel.com/test_sets/ca53ff1e-4c71-11e6-9f8e-5254006e85c2