Bug 2015543

Which OSP version is affected? ... how to reproduce?

(In reply to Matthias Runge from comment #1)
> Which OSP version is affected? ... how to reproduce?

I have no additional information, I've only created this bug because attached customer case has been added to already close RHV bug.
Rohit, as it's your case could you please reply?

This is interesting from RHV point of view, as always uses latest libvirt from AV for EL8 and we still support sending data from RHV hypervisors through collectd to external ElasticSearch

Hello Matthias,

As per customers update-

No specific steps to reproduce. It just happens after deployment.
This is the guide I follow to enable STF.
https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.1/html/service_telemetry_framework_1.3/index

It's OSP 16.1.


Here are some more details:

()[root@compute /]# virsh dommemstat instance-00000XYZ
actual 524288
swap_in 0
swap_out 0
major_fault 0
minor_fault 74765
unused 441644
available 489032
usable 459116
last_update 1634605069
disk_caches 19096
hugetlb_pgalloc 0
hugetlb_pgfail 0
rss 270968

https://github.com/collectd/collectd/blob/main/src/virt.c#L940
static const char *tags[] = {"swap_in",        "swap_out",   "major_fault",
                               "minor_fault",    "unused",     "available",
                               "actual_balloon", "rss",        "usable",
                               "last_update",    "disk_caches"};

Those 2 hugepage stats is not supported by collectd.

Thank you, I found the api change for libvirt, see https://libvirt.org/html/libvirt-libvirt-domain.html#virDomainMemoryStatTags
However, I don't have insight where this version is deployed to.
It looks like we need to fix 16.1+ (all upper versions).

That looks like the following quick and dirty (and also untested) patch should fix the issue

diff --git a/src/virt.c b/src/virt.c
@@ -937,10 +951,11 @@ static void memory_submit(virDomainPtr dom, gauge_t value) {
 
 static void memory_stats_submit(gauge_t value, virDomainPtr dom,
                                 int tag_index) {
-  static const char *tags[] = {"swap_in",        "swap_out",   "major_fault",
-                               "minor_fault",    "unused",     "available",
-                               "actual_balloon", "rss",        "usable",
-                               "last_update",    "disk_caches"};
+  static const char *tags[] = {
+      "swap_in",        "swap_out",    "major_fault",    "minor_fault",
+      "unused",         "available",   "actual_balloon", "rss",
+      "usable",         "last_update", "disk_caches",    "hugetlb_pgalloc",
+      "hugetlb_pgalloc"};

in order to reproduce the issue, you'll need to add 
Extrastats "memory"
to the virt plugin configuration.

Hello,

"memory" is listed by default, no need to add it specifically.
https://github.com/openstack/tripleo-heat-templates/blob/master/deployment/metrics/collectd-container-puppet.yaml#L473

*** Bug 2038881 has been marked as a duplicate of this bug. ***

moving this to modified, since the build is available.

Hello,
Our Customer is asking for an update of the case associated to this BUG.
About this request, is there any vision when the fix planned will be available also in RHOSP 16.2 ?

Thank you so much in advance

No "Array index out of bounds" message in collectd.log on compute nodes. OSP16.1z9

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Red Hat OpenStack Platform 16.1.9 bug fix and enhancement advisory), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2022:8795