This release fixes several bugs, and improves the efficiency of the shared memory counters framework.
If you have many dynamic backends with various TTLs, many counters will be created and retired. This use case was previously not optimized, and tools like varnishncsa, varnishlog and varnishstat saw a lot of CPU usage. This has been fixed, but you may experience that it can take up to one second before the new counters appear in the varnishstat interface. This time is configurable through the parameter vsm_publish_interval.
I addition to minor fixes in MSE, VHA 6 and Edgestash, the following VMODs have seen minor fixes/updates:
See Changelog for more details.