Troubleshooting
Problem
DB2CLUSTER -CLEAR -ALERT FAILS TO CLEAR EXISTING ALERT FILES ON THE INSTANCE SHARED FILE SYSTEM
Symptom
At first , we may find the /var directory fulled , db2instance -list report ALERT , and we can't connect to that problematic node .
db2cluster -list -alert
1.
Alert: Host name xxxxx has 0 MB free space available on /var file system. Failure to write to /var due to no space will result in DB2 cluster failure.
Action: The file system requires a minimum of 25 MB free disk space be available. Free up space on /var file system. The alert will be cleared automatically when a suf
ficient amount of space becomes available on the filesystem.
Impact: DB2 Cluster Services may not function properly on the specified host and may eventually lead to a DB2 instance failure.
df -h
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/vg00-lv04 20G 3.9G 15G 21% /var
While , after remove unnecessary files and free up the space , the alert still exists , the connection to that node is abnormal , and also can't clear up the alerts by "db2cluster -clear -alert "
Cause
"db2cluster -clear -alert" fails to remove existing alert files from the instance's shared file system. This causes "db2instance -list" and "db2cluster -list -alert" to report that alerts exist within the cluster, even after the alert conditions have been satisfied. We opened APAR IT14504(included in V10.5fp8) intended to add functionality to "db2cluster -clear -alert" to delete any existing alert files from the file system in order to clean up the alerts.
Environment
pureScale environment
Resolving The Problem
1. Manually remove the alert files (*.fsalert) from the instance shared file system directory, sqllib_shared/ctrlha/alert.
2. Request special build for APAR IT14504
Was this topic helpful?
Document Information
Modified date:
16 June 2018
UID
swg21989539