MSDP DiskPools Problems..

[Spartacus81]

Friends,

enviornment NBU AIR running SuSe Linux 10SP2 NBU 7.5.0.4 and spoold EEB 2964167

Backups failing on both diskpool (source and target domain), including all other taks that NBU AIR does (replication, import etc bidirectional) i had the problem before where i noticed on one of the media servers the databases area were 100% full and CRDBDUMP folder eating almost 70% of the space.. noticed there were number of files with different dates so removing the old files did the trick; and crcontrol --getmode command showed YES; ran the cleanup job and CRQP to make sure dedup engine does what it normally do. 

NOW database area is full again CRDBDUMP is the culprit. i have ran the CRQP and its showing both YES (pending and busy) but i cant run backup.. i know you will think check the dam diskspace but there is plenty diskpool is only 50% full but perhaps its suggesting /databases area is full so free up some space there.. .. storaged.log and spoold.log also saying check the diskspace and there is nice little line saying CRQP has failed 5 times please contact support immediately!  

i have 20 media servers (10 at each site) in the enviornment with exactly the same spec and storage attached to it but there are only two that are pain in the bum at the moment.. looking at the databases area on other MSDP they are less than 30% full.. 

can someone please shed some light that where the problem is and how can this be fixed..  

Problem:

Diskpool is up but /databases area is 100% full.. compaction on --getmode showing GET, PUT, STORAGED all YES

CRQP is running both pending and busy YES.. why CRDBDUMP is getting full and how can this be rectified? 

as always everyones input and knowledge shared will be appreciated.

[Spartacus81 · Accepted solution]

Apologies for the late update if anyone has been following it.. 

the problem was Symantec suggests to have a 15% space of /data partition allocated to /databases partition... for example: if your /data partition size is 20TB you should allocate minimum 3TB space to database partition..

to me thats a complete waste of space however after speaking to backline engineers they suggests 1TB should be enough as i have 20 other media servers with 300GB allocated to database and thry are running fine.

Basically the problem i was running into  when the CRQP runs it doubles the size of CRDBDUMP area which lead to database partition full condition hence spoold crashes and marked diskpool down.. pointing CRDBDUMP area to another location resolved the problem,.