25 Mar 2018 |
@aparcar:matrix.org | Bmx runs on WiFi, I turn of WiFi with a switch, bmx crashes and stuck until restart | 17:27:24 |
@aparcar:matrix.org | That happend to me, was wondering of that happened to you as well | 17:27:53 |
nicoechaniz | if you check the video, bmx6 is working on oncelotes-bbone | 17:28:25 |
nicoechaniz | or so it seems from it's bmx6 -c ... outputs | 17:28:42 |
nicoechaniz | it has not crashed | 17:28:47 |
nicoechaniz | but the next node (nicojesigioia) is out of the bmx6 mesh. When we restart bmx6 in the oncelotes-bbone node, nicojesigioia starts meshing again | 17:29:24 |
nicoechaniz | I have also seen what you mention. But this is a different issue | 17:29:39 |
axn | Thats strange, bmx should fully recover approx 100 seconds after a node restarts. That is the timeout to purge state of stale nodes | 22:23:46 |
26 Mar 2018 |
axn | Btw, nicoechaniz did you check for bmx6 log messages. They should report about blocked-nodes reasons. Could not see your video yet but asap... | 15:59:04 |
27 Mar 2018 |
@aparcar:matrix.org | axn: bmx7 just crashed on two 841 machines, can I find logs anywhere? | 09:53:24 |
axn | Logread or dmesg? Maybe there is a core left? | 17:42:36 |
28 Mar 2018 |
axn | nicoechaniz: I watched your video again, but it does not show enough information. Please Avoid restarting nodes, this will always confuse neighboring nodes to block them temporally. Instead, if issue occurs, check nodes along the path with 'bmx6 -cd8'. If one is blocked its showN in originators section. Then check bmx6 debug messages for issues. They should appear in syslog, also you can always get them live with 'bmx6 -cd0'. Just leave the logging running during such incident. If you find something send cd0 and cd8 outputs. If that all doesnt lead forward we can try a remote pair-debugging session and try to debug your case systematically. | 06:09:43 |
@aparcar:matrix.org | axn: aaand bmx7 crashed again without no info in logread nor dmesg | 12:11:49 |
@aparcar:matrix.org | how can I find information? | 12:11:56 |
@aparcar:matrix.org | axn | 13:29:15 |
@aparcar:matrix.org | Wed Mar 28 12:09:24 2018 user.notice bmx7-auto-gw: We got Internet access, announcing it to the mesh
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: INFO kernel_dev_tun_del(): closing tunnel dev=X7C4main
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: INFO kernel_dev_tun_del(): closing tunnel dev=X7C6main
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: WARN dev_deactivate(): deactivating dev=wlan0-mesh_13 llocal=fe80::ea94:f6ff:feb6:9eef global=fd70:cf27:2acb:ddad:e4d3:963b:e4a4:d543
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: WARN dev_deactivate(): deactivating dev=eth0.2 llocal=fe80::ea94:f6ff:feb6:9eee global=fd70:cf27:2acb:ddad:e4d3:963b:e4a4:d543
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: WARN dev_deactivate(): deactivating dev=wlan1-mesh_13 llocal=fe80::ea94:f6ff:feb6:9ef0 global=fd70:cf27:2acb:ddad:e4d3:963b:e4a4:d543
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: WARN dev_deactivate(): deactivating dev=br-lan llocal=fe80::ea94:f6ff:feb6:9eee global=fd70:cf27:2acb:ddad:e4d3:963b:e4a4:d543
Wed Mar 28 12:18:27 2018 daemon.err bmx7[7466]: ERROR : raising SIGSEGV again ...
| 13:29:22 |
axn | Are you able to reproduce the crash? Is there a core in /tmp ? | 16:28:17 |
30 Mar 2018 |
nicoechaniz | @axn:matrix.org: OK, I will do that next time. | 00:31:14 |
1 Apr 2018 |
@aparcar:matrix.org | axn: where do you set the entry for /etc/sysupgrade.conf to preserve /etc/bmx7. I can't find it in the code. Can this be related? https://github.com/bmx-routing/bmx7/blob/643976ebfea9e67061c2c3666f17880a9115cc2b/sec.h#L27 | 12:32:18 |
axn | Normally you specify via openwrt makefile. Thought this line should do it: https://github.com/openwrt-routing/packages/blob/master/bmx7/Makefile#L143 but check also the lines 131ff | 22:05:28 |
2 Apr 2018 |
@aparcar:matrix.org | thanks | 05:39:05 |
@aparcar:matrix.org | how come there is no debian package for bmx7. would it work or is the overhead to complicated? | 05:39:23 |
@aparcar:matrix.org | axn: can I determina with the bmx7 command or even better from the json files if the current node works as a bmx7 gateway? | 06:08:41 |
3 Apr 2018 |
axn | aparcar: Nothing speaks against debian package. Libz and libmbedtls needed as dependency. Just for bmx7_config.so plugin there is the problem of a missing uci debian package and for bmx7_iwinfo the missing iwinfo debian package, but support for that plugins would be optional. | 07:32:43 |
axn | Any node that specified tunIn operates as a GW. That could of course be determined with 'bmx7 -c parameters' or from /var/run/bmx7/json/parameters | 07:35:59 |
axn | In practice, any node using bmx7 tunnels is also operating in GW mode. Some more and some less :-). Read more here: https://github.com/bmx-routing/bmx7#tunnel-announcements | 07:37:46 |
axn | aparcar: > aaand bmx7 crashed again without no info in logread nor dmesg Think I catched the bug you hit: Got introduced recently via d72b284816504aaae9712de4b4f756268b35a536 (so not yet included via openwrt-routing-packages) and fixed via 4508eadf1ee088396f9f762582bce80bc85e021c | 10:05:48 |
14 Apr 2018 |
nicoechaniz | hi @axn just wanted to report a bmx6 crash that is occuring a couple times a day on the same node. This is a logread portion | 18:28:59 |
nicoechaniz | https://pastebin.com/Qgvu5mnL | 18:29:04 |
16 Apr 2018 |
@aparcar:matrix.org | axn: please approve https://github.com/openwrt/packages/pull/5923 | 08:52:45 |