Precaution for Disk Full Situations of the log volume

 We have heard all of our life that precaution is better than cure ? Isn't it ?



A well know scenario which we face sometimes in HANA is log volume gets full i.e. the path in which log segments are stored is full and we don't have any additional space so that the DB can further write any new logs  , and we have many reasons for that to happen . Once we face the scenario of Log Volume full, Database gets into a hung state as Database has no way to write transaction log because even creation of new session would require writing some logs . Recovering from that would be in a different blog , but yes this is a very lengthy process. This blog is moreover like a small precaution which can give us a small time frame to come up with a solution in time when our HANA DB is still up and running and we can perform ALTER SYSTEM RECLAIM LOG

A very simple solution would be to have a monitor over log volume . 

Another trick would be to place a dummy file in the log volume which can be deleted when the log volume is full giving some space to Database, and some time to us so that we can RECLAIM LOG.

Command to create a dummy file in HANA Log volume 
dd if=/dev/zero of=<log_path>/dummy_allocator bs=2G count=25 

"if" represents the input file and "of" represents the output file so that exact copy of /dev/zero will be available in <log_path>/dummy_allocator which is just a dummy file. bs represents the block size which in this case is 2GB and count is number of block size which is 25 summing up to 50GB (File should give sufficient space for say 30 minutes of working time)



Read More : DD command

Comments