Got several reports of failures:
cpuset_memory_spread 5 TFAIL: hog the memory on the unexpected node(FilePages_For_Nodes(KB): _0: 2208 _1: 104132, Expect Nodes: 1).
and
cpuset_memory_spread 7 TFAIL: hog the memory on the unexpected node(FilePages_For_Nodes(KB): _0: 2432 _1: 104744, Expect Nodes: 1).
and
cpuset_memory_spread 13 TFAIL: hog the memory on the unexpected node(FilePages_For_Nodes(KB): _0: 103992 _1: 2100, Expect Nodes: 0).
A quick search and I found Ubuntu had been tracking similar failures for several years and since at least v5.0 kernel.
They didn't capture the exact failure string for each instance, but for the ones they did, it seems similar to the above, in that numbers on the "non active" node were between 2000 and 2500 - and hence fail the thresholds in the test:
/
In general, the cache hog will use more than 10000 kb slab space on the nodes
on which it is running. The other nodes' slab space has littler change.(less than 1000 kb).
/
upperlimit=10000
lowerlimit=2000
This is similar to issue #932 in that a combination of debug options and/or overall growth (bloat) since the test was introduced in 2009 has made the original quasi-arbitrary limit set back then too low for today.
Got several reports of failures: cpuset_memory_spread 5 TFAIL: hog the memory on the unexpected node(FilePages_For_Nodes(KB): _0: 2208 _1: 104132, Expect Nodes: 1). and cpuset_memory_spread 7 TFAIL: hog the memory on the unexpected node(FilePages_For_Nodes(KB): _0: 2432 _1: 104744, Expect Nodes: 1). and cpuset_memory_spread 13 TFAIL: hog the memory on the unexpected node(FilePages_For_Nodes(KB): _0: 103992 _1: 2100, Expect Nodes: 0).
A quick search and I found Ubuntu had been tracking similar failures for several years and since at least v5.0 kernel.
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1830585
They didn't capture the exact failure string for each instance, but for the ones they did, it seems similar to the above, in that numbers on the "non active" node were between 2000 and 2500 - and hence fail the thresholds in the test:
/ In general, the cache hog will use more than 10000 kb slab space on the nodes on which it is running. The other nodes' slab space has littler change.(less than 1000 kb). / upperlimit=10000 lowerlimit=2000
This is similar to issue #932 in that a combination of debug options and/or overall growth (bloat) since the test was introduced in 2009 has made the original quasi-arbitrary limit set back then too low for today.