we have a Virtual-Chassis containing 3 EX4300 (no mixed, no multi-gigabit) connected by the QSFP-Ports.
The Virtual-Chassis is preprovisioned. The QSFP-Ports are in default-state, which mean they are all intened to be used for building a VC.
Now we have the 2 issues:
By doing a show system partitions media internal, the system shows a "error: /dev/da0s1a is not a JUNOS snapshot" on the fpc0.
Also when doing a show system storage partitions, the system has mixed up da0s1a and da0s2a:
fpc0:--------------------------------------------------------------------------Boot Media: internal (da0)Active Partition: da0s2aBackup Partition: da0s1aCurrently booted from: active (da0s2a)
JATC recommend to do a format install on fpc0 to get the system back into normal state.
Can you please confirm the follwoing procedure:
Is this procedure correct ?
To answer to your questions:
1) "error: /dev/da0s1a is not a JUNOS snapshot" is related to PR1297888: When EX4300 is booting, "ERROR: /dev/da0s1a is not a JUNOS snapshot" could be seen from console logs, This is cosmetic error message and can be ignored as no impact.
2) If switch is booting from da0s2a is because da0s1a is corrupted and this is why a format install is required.
3) To format install EX4300 you can follow this guide: https://kb.juniper.net/InfoCenter/index?page=content&id=KB20643
As you mention, member that has file system corruption needs to be isolated and formated as standalone switch. The rest of the procedure is correct.
Greetings, Just to add to Jose's nice write-up, the format install will re-write the entire file system hence you will lose the configuration file, in case you are going to perform it on the master ( the RE) please make sure you have a backup configuration handy.
Regards,Dexter Cole .
Costa Rica Pura Birra! Peace & LoveJNCIE-ENT#863, 3X JNCIP-[SP-ENT-DC], 4X JNCIA [cloud-DevOps-Junos-Design], Champions Ingenius, SSYB
If this solves your problem, please mark this post as "Accepted Solution" so we can help others too \:)/
Hi Dexter, hi Joe
and thanks a lot for your input.
Absolutely, I will have a backup-config next to me when I will perform these steps 🙂
But I was thinking, that the re-integration of the node, even if the node was the former master should not lead to a situation where I have to push the backup-config into the VC.
Normally, if I will isolate the defetive node, in this case the master, the master-role should swap to the backup-routing-engine.
So if I take out the fpc0, make a format-install on fpc0, connect it back to the VC, I would assume, that the backup-routing-engine, which holds the master role will push the productive config to the fpc0 ... am I wrong - please feel free to correct me 🙂
Hi Mawoh,As you mentioned at the moment you remove the master, the backup member should take over, once you have completed the format install you can rejoin the member to the stack and configuraiton should be there.
In fact my advise will to perform the mastership switch before removing the master,>request chassis routing-engine master switch
You are correct, once you add fpc0 back, master will push configuration into backup RE.
If this solves your problem, please mark this post as "Accepted Solution"
You are welcome, something else you can do to avoid any unpleasant surprise, is to enable commit synchronize (if you have not) before swapping out the faculty member, which this you will rest assured that the back up with have the same exact configuration file as the backup. i.e
If this solves your problem, please mark this post as "Accepted Solution" so we can help others too \/
Great guys, thanks a lot for all of your great suggestions and input, highly appreciate this 🙂
One more question:
With which config should I then re-attach the node into the VC - atfer the format install is complete and the partion-structure is re-organized?
I guess I should use the factory default config (with no root-pw set ?) -> correct ?
PS: I have also ordered 3xEX4200 (3x EX4300 would have become too expensive 🙂 ) for my private lab - I will lab this whole recovery-scenario in advance before doing this in production.
I was pleasantly surprised on how cheap the EX4200 ( I know it is quite an old model) are able to order (of course as an used model).
The advantage is that the EX4200 also have dedicated VC-Ports and are also able to run the same FW 14.1X53 as the EX4300.
No configuration is needed as once you add backup RE it will pull the configuration from the Master.
As you mention, it will use the factory default config with no root-password set but once you add backup RE it with get the config from the current master.
Hope this clarifies your question.
You are more than welcome, glad to see that you could get some good deals with those EX4200's at first you might have thought that you were getting some trinkets but trust me they are far from that. I am almost sure that this is because of the EoL announcement for these devices, in early Feb they announced that they will be end of life & support in 2024 now we can see how this affects the price 🙂
PS: Please note that the syntax will be different from one device to the other EX4300's use the ELS software and the EX4200's don't. I attached a file aka cheat sheet that I found online about this
and sorry for the delayed answer - I currently have a couple of days off.
Thank you all so much for all of your support in that case - you guys are great !
@Dexter: Thanks for your cheat-sheet.
And yeah, I am absolut fan of junos (for serveral years already - Most of the time having fun with the great SRX-Boxes ).
So, thanks again to all of you - questions answered - post will be marked as solved.