Allow CAS (rsrv) to log dropped updates

Bug #1536987 reported by Ralph Lange
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
EPICS Base
Triaged
Wishlist
Ralph Lange

Bug Description

When CA clients are not able to keep up with the updates that the database pushes to CAS, CAS drops updates (overwriting the last item in the queue between database and TCP sending thread).

The fact that updates are dropped (i.e. being missed by the client) is not communicated.
The best option would be informing the client (in form of a CA message meaning "<n> updates missed") - but CA does not implement this.
The second best option would be logging the fact that updates were lost (dropped updates are already counted), so that some post-mortem analysis tool or an intelligent logging system (think Splunk) can be made aware of the problem.

This logging should be configurable (default=off) to avoid clogging up logfiles in installations that do not want to use the feature.

Revision history for this message
mdavidsaver (mdavidsaver) wrote :

This seems reasonable in concept, but cries out for granularity. Actually it occurs to me that this should perhaps be driven by the CA client end. Perhaps some "magic" filter to set a flag in the associated dbEventSubscription?

Changed in epics-base:
status: New → Triaged
importance: Undecided → Wishlist
assignee: nobody → Ralph Lange (ralph-lange)
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.