gdplogd losing connection to the router. signature write failure: Success
The relevant error message in the logs (running at *=20)
*** Processing command 70=CMD_READ from socket 10 _gdp_req_dispatch >>> CMD_READ (70) [gcl->refcnt 1] disk_read(1): EOF _gdp_req_dispatch <<< CMD_READ ERROR: 404 not found [Berkeley:Swarm-GDP:404] _gdp_pdu_out, fd = 10, basemd = (nil): NAK_C_NOTFOUND _gdp_pdu_out: signature write failure: Success _gdp_event_cb: got EOF, 0 bytes left Re-using channel @ 0x1428290 process = 0x411287 advertise = 0x40613b Trying Zeroconf:
Followed by this, the log-server just starts looking for zeroconf. I am not sure if the 'signature write failure: Success' message is related to the lost connection, or if it is a separate issue. See attached log for one full run of gdplogd.
#5 Updated by Nitesh Mor about 4 years ago
Here's how to reproduce this:
- Use the visualization tool at http://gdp.cs.berkeley.edu:8888/ and go to Chair 1,
- Hit 'Plot'.
- Look at the system log of
gdplogd. It should be something like:
get_open_handle: opening OSbcXbklH_LEH2qqdo_PiiohaEAKJlf6fx6MQBuO934 _gdp_pdu_out: signature write failure: Success _gdp_event_cb: got EOF, 0 bytes left Trying Zeroconf:
OSbcXbklH_LEH2qqdo_PiiohaEAKJlf6fx6MQBuO934 maps to
edu.berkeley.eecs.swarmlab.device.c098e530002b, which exists but is empty.