Hi, Remi, our issue happen again even we reduce the RPC call.
Also, I notice when use “eventlet”, when we download a 8M log from web, the chrome will reported “network error”, and the server log reported: {"log":"[Mon Sep 28 06:56:27.101163
2020] [proxy_http:error] [pid 189:tid 140256774096640] (70008)Partial results are valid but processing is incomplete: [client 10.192.244.206:56938] AH01110: error reading response,
Looks above similar to our job submit issue?
Then, I switch to use “sync” mode, and find now I can download the 8M log completely.
If possible this be an environment issue or? Yes, we have apache to proxy the request to gunicorn, but looks this is setup by lava release? (We use docker solution)
From: Remi Duraffort <remi.duraffort@linaro.org>
Sent: Friday, September 25, 2020 3:28 PM
To: Larry Shen <larry.shen@nxp.com>
Cc: Milosz Wasilewski <milosz.wasilewski@linaro.org>; lava-users@lists.lavasoftware.org
Subject: Re: [Lava-users] [EXT] Re: Issues about XMLRPC & lavacli.
Caution:
EXT Email
Le ven. 25 sept. 2020 à 03:50, Larry Shen <larry.shen@nxp.com> a écrit :
Hi, Remi, thanks.
>>> 1. the zmq event stream is still available.
Q: This means still a zeromq for lava-publisher, and only for lava-publisher?
zmq is still used internally to send messages from lava-scheduler and lava-server-gunicorn to lava-publisher. lava-publisher will then publish events via zmq or websockets.
>>> 2. It's also possible to use web sockets to get the event stream.
Q: Any documentation we can refer to so that we could had a try for this later?
See lavacli source code. Using python3-aiohttp that fairly simple:
Rgds
--
Rémi Duraffort
LAVA Architect
Linaro