To investigate deeper what is going on here
we need more data on why caching is failing, like obj.ttl, obj.hits, CDN server-host and whether there was a HIT/MISS/ERROR. That data should be included in the log one can download on the Squidex dashboard.
@Sebastian just as a reminder for when you get to this: you can check our DMs regarding that topic for more details.
It has actually been implemented.
1 Like
@Sebastian Nice, many thanks!
However, currently all requests seem to MISS
the cache and they all also have a TTL of 0.
I have migrated the squidex website to use CDN. Then I can give you more details next week.
1 Like
I am planing to log all requests. Logging a request means that it will be charged.
So far only the cache HITS have been tracked at the CDN and have been invoiced with a factor of 0.5. But this leads to problems with traceability.
Therefore the plan is as followed:
Reduce the factor to 0.25 and log all requests
Before (no shielding, partial logging)
CACHE MISS: Item not cached in edge server
CACHE HIT: Item cached in edge server
Edge: 0
Backend: 1
Total: 2
After
CACHE MISS: Item not cached in edge server and shield server
Edge: 0.25
Shield: 0.25
Backend: 1
==> Costs 2
PARTIAL CACHE HIT: Item not cached in edge server, but in shield server
Request (CACHE HIT): Item cached in edge server
This means:
CACHE MISSES become more expensive.
CACHE HITS become less expensive
=> The higher the cache rate, the lower the costs.
1 Like
Sounds like a very reasonable approach, looking forward to it!
system
Closed
August 31, 2021, 12:31pm
7
This topic was automatically closed after 2 days. New replies are no longer allowed.