Unsolved
This post is more than 5 years old
53 Posts
0
25283
Garbage Collection command line
Any one, knows the Command LIne to perfom a Garbage Collection manually
When the maintenance services is up the Garbage Collection run all the time the Blackout Window is working
But when the Garbage Collection is perform by GUI, its only one hour duration, So now I have 1 week, with new node, and the balance process is quite slow, in this scenario I will take 20 days in balance process. So when I run a Garbage Collection is running just one hour and its not enough for me.
I would like to know if any one knows how can I execute a Garbage Collection manually and change the value of 3600 seconds in order to get a better Garbage Collection process.
Regards.
rob_steele
120 Posts
1
November 5th, 2012 08:00
David, can you output capacity.sh on here ? I would also increase the blackout window and let the maintenance take care of the garbage collection automatically. You shouldn't have to manually run GC.
rob_steele
120 Posts
0
November 6th, 2012 07:00
So David, it looks exactly as I expected, you must have some jobs set to run in overtime and are not allowing an automatic garbage collect on a daily basis. Do you have very long running jobs ? I think our first step in this process is to figure out why GC isnt running daily.
dwilches
53 Posts
0
November 6th, 2012 07:00
And this is one day when the GC ran automatically
Date New Data #BU Removed #GC Net Change
---------- ---------- ----- ---------- ----- ----------
2012-10-26 429999 mb 527 -797525 mb 1 -367525 mb
As you see the difference is big, when you hace enough hours for GC process instead only one hour when is manually.
dwilches
53 Posts
0
November 6th, 2012 07:00
When a new node is added, the next step is the balancing process, In this way, you must to stop maintenance services. That´s the reason I perfom my GC manually instead automatically as when the maintenance service is running.
This is my output of capacity.sh
Date New Data #BU Removed #GC Net Change
---------- ---------- ----- ---------- ----- ----------
2012-10-28 432342 mb 518 0 mb 432342 mb
2012-10-29 480548 mb 562 0 mb 480548 mb
2012-10-30 387971 mb 525 0 mb 387971 mb
2012-10-31 441714 mb 527 0 mb 441714 mb
2012-11-01 588283 mb 529 0 mb 588283 mb
2012-11-02 376963 mb 536 -236819 mb 2 140144 mb
2012-11-03 454712 mb 536 -121603 mb 1 333109 mb
2012-11-04 453599 mb 538 0 mb 453599 mb
2012-11-05 473995 mb 592 0 mb 473995 mb
2012-11-06 307040 mb 489 0 mb 307040 mb
dwilches
53 Posts
0
November 6th, 2012 08:00
I have answer your question
The GC is not running automatically, because the maintenance service is down, And is down, because when you add a new node at your grid, you need to balance the size of nodes In this case, the maintenance activities do not have to run.
Scenario
My Blackout Window and Maintenance Window is down
This is my dpnctl status and as you see the Maintenance windows scheduler is suspended.
Identity added: /home/dpn/.ssh/dpnid (/home/dpn/.ssh/dpnid)
dpnctl: INFO: gsan status: degraded
dpnctl: INFO: MCS status: up.
dpnctl: INFO: EMS status: up.
dpnctl: INFO: Backup scheduler status: up.
dpnctl: INFO: dtlt status: up.
dpnctl: INFO: axionfs status: up.
dpnctl: INFO: Maintenance windows scheduler status: suspended.
When you have this situation, you need to run manually, Checkpoint, Validation, and Garbage.
The problem is, Garbage collection ru only one hour when is perfom by GUI. Got it?
dwilches
53 Posts
0
November 7th, 2012 15:00
I got this command line
avmaint garbagecollect --ava --kill=10 --usehistory=false --maxtime=9600
Anyone knows if this command line can be useful for my porpuse?
ionthegeek
2K Posts
1
November 7th, 2012 16:00
The idea behind disabling the maintenance scheduler during balancing is that running maintenance activities during balancing can cause issues on the system so maintenance should not run.
Running garbage collection while balancing is running on the system (or any maintenance activity, for that matter) is not a good idea. If maintenance needs to run, balancing on the system should be suspended and the maintenance scheduler re-enabled to allow a scheduled maintenance cycle to complete. The implementer responsible for your node add should be monitoring the status of balancing and managing this for you.
Note that the information above does not apply to Avamar 6.1. Balancing is now a maintenance task in Avamar 6.1 (meaning on Avamar 6.1 systems, maintenance should be enabled to make sure that the system runs balancing each day during the maintenance window).
20 days to balance is unusually long. In my experience, balancing generally completes in less than a week. How did you determine that it will take 20 days to complete the balance?
Finally, I would urge you not to run maintenance manually on any system. Running maintenance manually has risks that are not obvious. Before starting any maintenance activity, the maintenance scheduler may take actions like loading or unloading the index caches. Running maintenance manually does not take these actions. If the index caches are loaded when hfscheck starts, the system could run out of memory and trigger a node offline issue, for example.