Attention is currently required from: fixeria.
pespin has posted comments on this change by fixeria. ( https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/38309?usp=email )
Change subject: s1gw: use the new counter name (out_pkt.forward.unmodified) ......................................................................
Patch Set 1:
(1 comment)
File s1gw/S1GW_Tests.ttcn:
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/38309/comment/cb8d85ec_d03f8... : PS1, Line 206: {name := mp_statsd_prefix & "ctr.s1ap.proxy.out_pkt.forward.unmodified.value", mtype := "c", min := 2, max := 2} Well, imho the counter simply says what I explained: A received (input, incoming) packet was forwarded.
S1GW_CTR_S1AP_PROXY_IN_PKT_REPLY_ERAB_SETUP_RSP looks confusing as if the input packet was a reply
I cannot find this counter in the metrics patch?
Alternatively, we could remove the IN/OUT classification?
So to me that main problem right now in the current code base of osmo-s1gw, is that the parsing of pkts is done at the same time while operating/mangling it. This causes several problems like not having the full set of ERABS before operating on them.
So the operation of the proxy would be imho: 1- I take an in_pkt, decode it generate some data structures that will be used later on (dictionaries, lists, etc.) 2- Operate on the parsed packet (FSMs, PFCP, etc.) 3- Generate an output packet if/whenever needed
So for me clearly we have 2 separate sets of counters in the input and output phases. And we need to count on both phases, since eg. an error output packet may be generated in phase 1 or in phase 2, so we need to track that.