[j-nsp] MX960 power supply stopped during ISSU

[Aaron Gould]

Thanks Tim, unsure what you mean by .... "do you have enough power supplies
to maintain N+1  at full tilt?"  

Thanks Jerry, ISSU upgrade went fine.... this time.  I've done 3 so far...
-1st MX960 ISSU upgrade in lab-960 was fine
-2nd MX960 ISSU upgrade had a glitch, I don't recall why, but it caused me
to have to do a stand-alone RE upgrade to the old master/new bu re... but I
got through it and all is well. (as you may know, stand-alone upgrade caused
me to have to remove, gres, nsr, nsb...re-enabled it afterwards)
-3rd MX960 ISSU, the one this thread is about, went good.  ...except 2
issues, that did *not* hender the issu from completing successfully, so just
want to be clear on that.
--- Issue 1 was the power supply I found down afterwards.  But now I'm
learning from you all and JTAC that the ISSU had nothing to do with this
PEM0 going offline and it was purely coincidental that it went offline
during the ISSU.  I asked the JTAC to tell me why the power supply went
offline.  Waiting on that answer.  He did say that there are normal
procedures for a power supply going offline to simply reseat it.  if it
comes back up, great.  If not, rma.  In my case, we did not reseat, but
instead, removed it, but in a known-good one, and took that removed bad one
to the lab, where it has been functioning fine since yesterday... to which
the jtac is calling it good.
--- Issue 2 was , during the issu upgrade process, there is a config check
apparently (details shown below... see starting at "Validating against
/config/juniper.conf.gz".... well, strangely it found an entry for
"ether-options" and aborted the ISSU.  This was weird, that et (40 gig)
interface et-11/1/1 was alive and well config'd like that, and passing
traffic in a functional lag bundle ae43... I removed the ether-options
command, before commiting, tried to put it back and it would *not* let me...
weird.  Then I looked on other et-x/x/x interfaces in other ae's like ae40,
ae41, etc... and those were all config'd with gigether-options... so I
likewise config'd et-11/1/1 with gigether-options, commited, restarted ISSU,
and all is well.

- Aaron

Details of the 3rd MX960 upgrade 2nd issue... but was easily overcome and
restarted issu and completed successfully.
.............................
Mounting junos-daemons-mx-x86-64-20181003.235426_builder_junos_174_r2_s1
Verified jsdn-x86-32-17.4R2-S1 signed by PackageProductionEc_2018 method
ECDSA256+SHA256
Mounting jsdn-x86-32-17.4R2-S1.2
Verified jsd-x86-32-17.4R2-S1 signed by PackageProductionEc_2018 method
ECDSA256+SHA256
Mounting jsd-x86-32-17.4R2-S1.2-jet-1
Verified jpfe-common-x86-32-20181003 signed by PackageProductionEc_2018
method ECDSA256+SHA256
Mounting jpfe-common-x86-32-20181003.235426_builder_junos_174_r2_s1
Verified jfirmware-x86-32-17.4R2-S1 signed by PackageProductionEc_2018
method ECDSA256+SHA256
Mounting jfirmware-x86-32-17.4R2-S1.2
Verified jdocs-x86-32-20181003 signed by PackageProductionEc_2018 method
ECDSA256+SHA256
Mounting jdocs-x86-32-20181003.235426_builder_junos_174_r2_s1
Hardware Database regeneration succeeded
Validating against /config/juniper.conf.gz
/config/juniper.conf:390:(21) syntax error at 'ether-options'
  [edit interfaces et-11/1/1]
    'ether-options {'
      syntax error
/config/juniper.conf:392:(9) error recovery ignores input until this point
at '}'
  [edit interfaces et-11/1/1]
    '}'
      error recovery ignores input until this point
mgd: warning: (#2)Db open failed...
Validation failed
ERROR:  Failed to add
/var/tmp/junos-vmhost-install-mx-x86-64-17.4R2-S1.2.tgz
da0 at vtscsi0 bus 0 scbus2 target 0 lun 0
da0: <QEMU QEMU HARDDISK 1.7.> detached
(da0:vtscsi0:0:0:0): Periph destroyed
error: ISSU Aborted!
[Jan 29 00:04:28]:ISSU: Error
Chassis ISSU Aborted
[Jan 29 00:04:28]:ISSU: IDLE
warning: Host software installation has failed.