Note: This bug is displayed in read-only format because
the product is no longer active in Red Hat Bugzilla.
RHEL Engineering is moving the tracking of its product development work on RHEL 6 through RHEL 9 to Red Hat Jira (issues.redhat.com). If you're a Red Hat customer, please continue to file support cases via the Red Hat customer portal. If you're not, please head to the "RHEL project" in Red Hat Jira and file new tickets here. Individual Bugzilla bugs in the statuses "NEW", "ASSIGNED", and "POST" are being migrated throughout September 2023. Bugs of Red Hat partners with an assigned Engineering Partner Manager (EPM) are migrated in late September as per pre-agreed dates. Bugs against components "kernel", "kernel-rt", and "kpatch" are only migrated if still in "NEW" or "ASSIGNED". If you cannot log in to RH Jira, please consult article #7032570. That failing, please send an e-mail to the RH Jira admins at rh-issues@redhat.com to troubleshoot your issue as a user management inquiry. The email creates a ServiceNow ticket with Red Hat. Individual Bugzilla bugs that are migrated will be moved to status "CLOSED", resolution "MIGRATED", and set with "MigratedToJIRA" in "Keywords". The link to the successor Jira issue will be found under "Links", have a little "two-footprint" icon next to it, and direct you to the "RHEL project" in Red Hat Jira (issue links are of type "https://issues.redhat.com/browse/RHEL-XXXX", where "X" is a digit). This same link will be available in a blue banner at the top of the page informing you that that bug has been migrated.
Description of problem:
`crm_mon -s` prints "CLUSTER OK" when there are unclean (online) nodes
The '-s' switch for `crm_mon` is meant to provide a one-line output suitable for Nagios. Currently, `crm_mon -s` prints "CLUSTER OK" unless one of the following conditions is met:
- no DC
- at least one offline node
If there is a node in "unclean (online)" status -- e.g., with a pending/failed fence event against it -- the output will still be "CLUSTER OK". This leads an administrator to believe that the cluster is completely healthy when in fact it is not, even though all members are online.
The '-s' switch causes the print_simple_status function to execute: https://github.com/ClusterLabs/pacemaker/blob/Pacemaker-1.1.18/tools/crm_mon.c#L884-L923
There are only two calls to the `mon_warn` macro: line 828 and line 841.
The `for` loop that processes node status does not handle "unclean" state in any special way, as the print_status function does. It only checks:
- standby
- maintenance
- online/offline
831 for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) {
832 node_t *node = (node_t *) gIter->data;
833
834 if (node->details->standby && node->details->online) {
835 nodes_standby++;
836 } else if (node->details->maintenance && node->details->online) {
837 nodes_maintenance++;
838 } else if (node->details->online) {
839 nodes_online++;
840 } else {
841 mon_warn(" offline node: %s", node->details->uname);
842 }
843 }
So a couple of approaches to this problem, if we want to print "CLUSTER WARN" in case of any unclean nodes, would be to:
- add a mon_warn() call in case of `node->details->unclean && node->details->online` and leave the rest as it is
- eliminate the "offline node" mon_warn() call in favor of "unclean (online)" and "unclean (offline)" mon_warn() calls
--------------------
Version-Release number of selected component (if applicable):
pacemaker-1.1.18-11.el7.x86_64
reproducible in master
--------------------
How reproducible:
always
--------------------
Steps to Reproduce:
1. Create a GFS2 Filesystem stack (controld, clvm, Filesystem) on a two-node cluster.
2. Remove the stonith device for node 2, or remove node 2 from pcmk_host_map.
3. Unmount the GFS2 filesystem on node 2, causing a monitor failure.
4. Run `crm_mon -1` and verify that node 2 is in "UNCLEAN (online)" state.
# crm_mon -1
Stack: corosync
Current DC: nwahl-rhel7-5-clus1 (version 1.1.18-11.el7-2b07d5c5a9) - partition with quorum
Last updated: Tue May 8 16:56:22 2018
Last change: Tue May 8 16:55:58 2018 by root via cibadmin on nwahl-rhel7-5-clus1
2 nodes configured
7 resources configured
Node nwahl-rhel7-5-clus2: UNCLEAN (online)
Online: [ nwahl-rhel7-5-clus1 ]
Active resources:
Clone Set: dlm-clone [dlm]
Started: [ nwahl-rhel7-5-clus1 nwahl-rhel7-5-clus2 ]
Clone Set: clvmd-clone [clvmd]
Started: [ nwahl-rhel7-5-clus1 nwahl-rhel7-5-clus2 ]
Clone Set: clusterfs-clone [clusterfs]
clusterfs (ocf::heartbeat:Filesystem): FAILED nwahl-rhel7-5-clus2
Started: [ nwahl-rhel7-5-clus1 ]
xvm (stonith:fence_xvm): Started nwahl-rhel7-5-clus1
Failed Actions:
* clusterfs_monitor_60000 on nwahl-rhel7-5-clus2 'not running' (7): call=26, status=complete, exitreason='',
last-rc-change='Tue May 8 16:55:40 2018', queued=0ms, exec=0ms
5. Run `crm_mon -s` and observe that it prints "CLUSTER OK".
# crm_mon -s
CLUSTER OK: 2 nodes online, 7 resources configured
--------------------
Actual results:
# crm_mon -s
CLUSTER OK: 2 nodes online, 7 resources configured
--------------------
Expected results:
# crm_mon -s
CLUSTER WARN: unclean (online) node: nwahl-rhel7-5-clus2
--------------------
Additional info:
N/A
Due to this not making the 7.7 time frame, I'm moving it to RHEL 8 only, as RHEL 7 will only be getting bug fixes from this point.
The approach I'm leaning to here is creating a new tool (maybe called crm_check) that would solely be a monitoring plugin, since it doesn't really overlap with crm_mon much. We would likely handle Bug 1577085 at the same time.
Unfortunately due to developer constraints, I cannot commit to a release time frame.
Upstream has deprecated the crm_mon -s option and will eventually drop it. Users are recommended to use a community-supplied or custom plugin instead.
If demand warrants, we could write a new plugin (separate from crm_mon) and submit it upstream. However at this time, especially given the diversity of monitoring systems in use (nagios, prometheus, zabbix, etc.), it seems unlikely to be a priority.