Is anyone out there successfully using cloverleaf 5.8 rev 3 with Unix 6.1.00? We have it here and seem to be having I/O issues…any input would be appreciated.
well we are on 5.8 Rev 3 unix os level 6100.05. When our nightly backup runs, all I/O is used and messages backup for 45 minutes….also when we run our cycle scripts, any scripts using a compress command, freezes the system, records backup in the engine and not delivered until scripts are done. Had IBM look at , and there seems to be something with file locks. Because of this, we had to comment out all compress and our backups are selective, until this is resolved. You will see, after you run your compress, i-nodes still created and space not freed up until process is cycled.
Yes, the cycle save and i-node issue was a known issue in an earlier post. You need to bounce the process to free up the i-nodes. You can probably search the list to find the exact post.
Thanks for the link…but in looking at the memory leak and smat file issues, it was suppose to be fixed in REV3… we are on REV3, unix os 6.1.0.0.5 and still experiencing this….we have ticket open with lawson and are awaiting response from R&D.
yes…we are on 5.8.3 and are still having issues….again, R&D is looking at code and we have weekly calls between them and IBM and ourselves. Will let you know what happens…hopefully soon…
We are planning to upgrade to CIS5.8 Rev3 as well on AIX6.1 TL 6 SP5
from CIS5.7R2 (may apply R3 here soon). My question is: have you enabled multi-threading for your engines (processes)? It seems from the posts and my informal talks at CUE11 with some Cloverleaf staff that most of the problems with the 5.8 engine center around the revampled core modules (C code) dealing with multi-threading. Apparently, the AIX6.1 platform is sensitive to these changes now versus other Unix platforms.
Please (!) post any response from Cloverleaf R & D concerning the CPU utilization problems that your shop is experiencing.
Thanks from this community member for keeping us in the loop.
Just a thought here but did you check the AIX installation instructions about adjusting the user MAXUPROC settings which includes number of open file handles? The recommendations is up to 3X your current values.
Maybe a red herring but doesn’t hurt to check them.
We did not notice on our TEST box either….on your test box after you run all you compress scripts, use the following command and see if you are left with I-node file
**note: i dont have any on my test box…just production** I dont have the volumn. Even with the compress gone from our scripts, we are still getting new inodes…
set you default to /healthvision (or whatever you call your directory)
fuser -dV /healthvision
CLOVERLEAFP as hci # > fuser -dV /healthvisionp
/healthvisionp:
fd=24 7602380
fd=21 7864456
fd=24 17629320
fd=21 19988508
fd=29 34341042
fd=21 38404236
fd=24 40632532
fd=21 41156640
fd=24 49938530
fd=22 55574548
fd=24 57475274
fd=33 59572252
***above all all the inodes created, i must stop and restart all the processes to clear up***
.
