Hello colleagues, seems I need some help here.
Scenario:
- 4 LAB QFX 5100 (2 spines and 2 leaves) simulating our production network need to be upgraded from 14.1x53 to qfx-5-18.1R3-S10.4 (yeah I know it's written somewhere you can't skip 3 version but it is not applicable here and we confirmed it with Juniper) anyway the question is not here.
TEST1
- I Built the VCF connected it with few switches to test ping losts and did the direct(1 step) upgrade for all four switches and all went ok.
TEST2
- I copied the production configuration to test and did the upgrade and it is failed, I repeated it 3 times and always failing!!
actually upgrade is ok as version is updated but, I am not able to commit the config.
difficult to understand how the configuration is affecting the upgrade but that what I got trying 3 times, below I put related logs.
- the way I fixed it each time is to zeroise all 4 switches one by one then copy the config and then it works, but when upgraded with the config it keep failing.
---------------------------------------------
mostly here is the error I get after upgrading with my production config when trying to commit
---------------------------------------------
root# commit check
error: Only one source address allowed in ntp for default routing-instance
error: configuration check-out failed: daemon file propagation failed
the first one because I have this in my configuration which working fine with 14.1x53
set system ntp source-address 10.x.x.x
set system ntp source-address 10.y.y.y
set system ntp source-address 10.z.z.z
so after deleting two and keeping one, this error is fixed but still commit is failing:
root# commit check
JNUD_CONFIG_ERROR: Configuration database has errors
error: configuration check-out failed
and here I have no other option the do request system zeroiseto restore it to factory default then copy the config manually and fix some errors related to VCF, NTP and it works.
So what I am missing here, we are planning to do that on our production network soon with small outage window and I don't go via the night mare of needing to zeroise all 10 vcf switches and configuring all from the scratch, I also don't want to delete the current config, do the upgrade then restore it, I want to understand how can I get the upgrade succeed with present configuration.
---------------------------------------------
After upgrade in two cases I got this:
---------------------------------------------
pci-hgcomdev module loadedhgcommdev0: <HGCOMMDEV For Host VM communication> mem 0xfebc1000-0xfebc1fff at device 22.0 on pci0
hgcommdev0: hgcommdev: registers at 0xf7b9e000
Creating initial configuration...Kernel out of physical pages (error count: 1)
mgd: error: Only one source address allowed in ntp for default routing-instance
mgd: error: commit failed: daemon file propagation failed
Warning: Commit failed, activating partial configuration.
Warning: Edit the router configuration to fix these errors.
override rw-r----- root/wheel for /var/etc/pam.conf? (y/n [n]