Skip to main content
Flexible Top Header
Question

Address Manager notification processing

  • December 19, 2025
  • 6 replies
  • 60 views

Forum|alt.badge.img+2

I’m curious if anyone can speak to what the limiting factors are on BAM’s ability to process notifications?  We’ve experienced some occurrences where it looks like the notification processing has hit a SQL block for a period of time, causing PNA file backlogs on the BAM.  Once the block clears, it takes BAM a certain amount of time to catch back up.  Observing the appliance while it does so, it seems a single postmaster process CPU-bound against a single thread - so depending on the size of the backlog it can take quite a bit of time to catch up, even though the BAM overall is 98% idle.

This notification processing is really the only performance-related concern I currently have for our BAMs - and it would be awesome if  notification processing could be improved in such a way as to better consume the available hardware resources.

6 replies

bshorland
Forum|alt.badge.img
  • BlueCat Employee
  • December 22, 2025

Hey Tim,

Any AAAA record updates occurring? There is a known performance issue with the processing of AAAA in particular that can cause a large slow down on the processing of the notifications, this has yet to be addressed

 

Brian


Forum|alt.badge.img+2
  • Author
  • Verified Resolver
  • December 22, 2025

Hi Brian - no, no AAAA updates.  Our PNA monitoring scripts (which alert if there are > 5 PNA files on BAM) have been triggering more often lately, and I’ve started looking into database optimization (we have about 700k orphaned MACs that need to be cleaned up) and perhaps a reindexing is due.  The past month or two we’ve had, usually on Monday mornings, a backlog of up to around 50-75 PNA files, but last week we went from 30 to over 350 in the space of an hour.  It took almost 24hrs for BAM to get caught up.  We don’t currently watch the notification-related SNMP OIDs so I’m not really sure if our volume of notifications spiked or something else occurred (there were some deadlock messages logged by postgres but nothing that lined up directly at this time).  For this reason, I really like the 25.1 notification-related stats now graphed in BAM so looking forward to getting there next year.  

 

But, as previously mentioned, it rankles a bit to see the BAM at overall ~95-98% idle with one thread pegged while it struggles to dig itself out of the hole all day.


bshorland
Forum|alt.badge.img
  • BlueCat Employee
  • December 23, 2025

Hi Tim, damn was hoping it was going to be something at least known … I would recommend in this instance getting a Care case opened up so Defenders & Pheonix teams can start looking at the root cause of what’s occurring, it could be as simple as the orphaned MACs slowing processing but those are supposed to be cleaned up during the audit data retention process automatically I thought since 9.3 days

Brian


Forum|alt.badge.img+2
  • Author
  • Verified Resolver
  • December 23, 2025

Thanks Brian.  I woke up just now to an alert for 584 PNA files so something is definitely amiss.  I’ll be opening that case shortly.  I already have a case open to deal with the 700k orphaned macs (my theory is that they exist pre-implementation of the 9.3 cleanup behavior and so I need to get to a “clean slate” first to allow for normal automatic maintenance).


Forum|alt.badge.img+2
  • Author
  • Verified Resolver
  • December 30, 2025

Follow on question for ​@bshorland or anyone else who may know - is the Notifications Per Second stat in an Integrity X BAM dashboard the number of Notifications received (and potentially queued) by BAM, or the number of notifications processed by BAM?  


bshorland
Forum|alt.badge.img
  • BlueCat Employee
  • December 30, 2025

AFAIK ​@tmaestas 

The notifications per second metric refers to the number of notifications processed by Address Manager per second. For reference, the upper limit for processing notifications on a BAM8500 is up to 700 notifications per second

Brian