slact
Cloudflare intermittency
When behind Cloudflare, there is intermittency in the messages getting across to the client. I see in the documentation that there is a hack for this using this parameter: nchan_subscriber_message_id_custom_etag_header I am not certain what value should be used for this parameter. Are they any other configuration changes needed to make nchan work consistently behind Cloudflare?
It appears Cloudflare wasn't an issue, it was using v1.2.6, reverting back to v1.2.5 to seem to resolve the intermittency.
Here is the error log for from 1.2.6:
2020/05/07 21:01:08 [alert] 61230#0: aborting
../nchan/src/store/spool.c:479: spool_fetch_msg: Assertion spool->msg_status == MSG_INVALID' failed. 2020/05/07 21:02:52 [alert] 78189#0: worker process 62628 exited on signal 6 ../nchan/src/store/spool.c:479: spool_fetch_msg: Assertion spool->msg_status == MSG_INVALID' failed.
2020/05/07 21:02:54 [alert] 78189#0: worker process 62753 exited on signal 6
2020/05/07 21:04:35 [error] 62627#0: MEMSTORE:05: tried adding WAITING chanhead 0000000001ACF3A0 excelerate/bbd3e4deede5cf2c971e52c45836f2b6118442345d63741b57b59247c7585dc1 to chanhead_gc. why?
2020/05/07 21:04:36 [error] 62629#0: MEMSTORE:07: tried adding WAITING chanhead 0000000001ACF3A0 excelerate/bbd3e4deede5cf2c971e52c45836f2b6118442345d63741b57b59247c7585dc1 to chanhead_gc. why?
../nchan/src/store/spool.c:479: spool_fetch_msg: Assertion `spool->msg_status == MSG_INVALID' failed.
2020/05/07 21:04:38 [alert] 78189#0: worker process 62754 exited on signal 6
seems to be a dupe of #534
reverting to 1.2.2 should fix it. Or you fork and revert the commit mentioned in #534
