With this work, there is a new built-in HTTP extension plugin that can be used to export OpenMonitoring / prometheus-based metrics.
An example output on a dev host:
$ curl -H "Authorization: Bearer $BEARER_TOKEN" http://localhost:8000/metrics/
# HELP server_bytes Number of bytes read into the server
# TYPE server_bytes counter
server_bytes{direction="tx"} 882
server_bytes{direction="rx"} 3107
# HELP server_metadata XRootD server metadata
# TYPE server_metadata gauge
server_metadata{instance="f4hp7ql65f.local:1095",site="bbockelm_laptop",version="v5.5.4.post 248"} 1680294742
I am told by @ivukotic it would be immensely helpful in "collapsing" the monitoring pipeline. His central Prometheus monitoring could work directly with the caches instead of going through the XRootD/UDP->TCP->Collector->LogStash->database (or in the case of OSG, XRootD/UDP->Shoveler/AMQ->Collector/AMQ->database).
Beyond shortening the chain it also simplifies the authorization scheme -- only the central prometheus instance needs a token. The caches just need to be configured to trust the central service.
Note this includes a dependency on https://github.com/jupp0r/prometheus-cpp; the build of the module should be disabled if the library is missing. This only needs the prometheus-cpp-core
library; it might be quite reasonable to think about vendoring it.
@jthiltges @djw8605 @biozit @brianhlin -- you guys might be interested in this as well.
https://github.com/xrootd/xrootd/pull/1985
(16 files)
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1