Quick update on this one...
this thread and another one were based on myself unable to deploy EX2300 switches due to 2 problems: loss of communication with the switches (PR1442376) and random swizzle reboot (not sure if there is a PR for this, but I have 2 Service request so far with Juniper)
PR1442376:
so 236 EX2300 switches running pre-release of 18.2R3S2.7 for 15 days now and no zombie switch. PR1442376 seems to be under control. Can't wait for the official release.
SWIZZLE REBOOT
as for the getty request that seems to induce swizzle reboot, I was told by Juniper to check for defective cables connected to the miniusb port. Problem is that no cables are connected, so I decided to just turn it off (set system aux port-type rj45 disable turn off the miniusb console but the rj45 console still works. Go figure ...) .
The trick of shutting dow the miniusb console ports seems to work.
Out of the 236 switches, I have 34 that issue getty requests on ttyu1 and on which I left the miniusb console port open to compare. all the other 200 switches had their miniusb port disabled and none of them swizzled reboot. but in the last 2 weeks. I am up to 13 of the 34 switches that went swizzle reboot during that time.
(how do I determine if a switch issue getty request ? message start messages for 60 secs and watch for
Oct 25 13:01:19 cslt-bearn-01-BYPASS init: getty repeating too quickly on port /dev/ttyu1, sleeping 30 secs
Oct 25 13:01:49 cslt-bearn-01-BYPASS getty[76714]: tcsetattr /dev/ttyu1: Invalid argument
Oct 25 13:01:50 cslt-bearn-01-BYPASS getty[76895]: tcsetattr /dev/ttyu1: Invalid argument
Oct 25 13:01:50 cslt-bearn-01-BYPASS getty[76896]: tcsetattr /dev/ttyu1: Invalid argument
Oct 25 13:01:50 cslt-bearn-01-BYPASS getty[76903]: tcsetattr /dev/ttyu1: Invalid argument
Oct 25 13:01:51 cslt-bearn-01-BYPASS getty[76904]: tcsetattr /dev/ttyu1: Invalid argument
Oct 25 13:01:51 cslt-bearn-01-BYPASS init: getty repeating too quickly on port /dev/ttyu1, sleeping 30 secs
so I will now turn off the miniusb on all my switches and I expect not to see a swizzle reboot anymore.
If that works, well, I seems to have a way to go to handle these 2 problems.
Juniper is working on the swizzle reboot thing. In the meantime, if the trick prevent a switch to randomly reboot, well, I'll take it !!!
Michel