we have a case that fsimage was missing on both namenode machines
this case is very rare , but in fact cluster we installed from scratch because that
so my quastion is is it necescery to backup these files ? ( for example each day by crontab job)
so in case we need to restore these files then we can take them from backup
It's always a best practice to backup the '/hadoop/hdfs/namenode/*' in very timely manner using crontab scheduler. FYI I'm doing the same keep backup for last 5 days by tar and moving them to different location(not on same server) for 12hours.
Hope this helps you.
in case of crash do you have procedure for restore , so I will test the restore on test system ?