PDRS relocation mode does not work. Similar messages can be found in logs:
23-09-14 17:03:17.871 open(/pcs/.drs/memory.node_list) : No such file or directory 23-09-14 17:03:17.872 command failed with error '2'
23-09-14 17:03:16.700 shaman-monitor(691656): Trying to evacuate the node 'md.cdf862e2e665320c' using the 'drs' mode... 23-09-14 17:03:17.872 shaman-monitor(691656): pdrs_schedule : command failed with error '2' 23-09-14 17:03:17.874 shaman-monitor(691656): /usr/share/shaman/pdrs_schedule exited with code 2
PRDS relocation mode does not work due to inability to find
/<pstorage_mount_mount>/.drs/memory.node_list file. This file contains information about all nodes in cluster joined into a shaman, specifically their current resources usage (license, ram, etc..). This data is used to decide a most suitable node for relocation.
File is generated by PRDS service with the use of SNMP protocol. Verify that all prerequisites are fulfilled:
Make sure necessary packages are installed:
# yum install shaman pdrs rmond
Make sure services are running:
# service snmpd status # service pdrsd status # service shamand status
Start services if any of them is stopped
iptablesto ensure SNMP traffic is not dropped
Make sure snmpd is properly configured
Snmpd configuration file,
/etc/snmp/snmpd.local.conf, should contain similar records:
rwcommunity parallels 127.0.0.1 .188.8.131.52.4.1.26171 rwcommunity parallels <IP_network/subnet_mask> .184.108.40.206.4.1.26171
IP_networkis the Parallels Cloud Storage cluster network address and
subnet_maskcovers all the nodes participating in the cluster.
shamandif changes to
# service snmpd restart # service pdrsd restart # service shamand restart
Tip! It might take a while to populate the file, give it some time. If after an hour file is not populated you may attempt to restart services again (on all nodes in cluster).