Syscom Todo: Difference between revisions

From CSCWiki
Jump to navigation Jump to search
No edit summary
No edit summary
 
(20 intermediate revisions by 7 users not shown)
Line 1: Line 1:
These are things that syscom should do eventually:
These are things that syscom should do eventually:

== Timeline ==
Get Mac Minis: Week of May 1

Get Intel Nucs: At least August but tentative

Get Rack Server: May

Get Netapp: Late May but tentative

Server Room Trip: May


==General==
==General==
* Prepare for `sodium-benzoate` upgrades/replacement.
* Prepare for `sodium-benzoate` decom.
** Need to find the physical server
** new-mirror has like 30 disk shelves so we can just do a live sync on the 2TB disks and then insert the 4TB ones
** Give to CSCF for decom
* Establish remote syslog
* Get UPS monitoring working across multiple systems
* `/users` backups
* `/users` backups
* Disaster recovery plan
* Disaster recovery plan
* Find out status of backup mirror
* Put backup containers onto cobalamin (auth2)
* Get an IP/KVM for the machine room which doesn't suck?
* Get an IP/KVM for the machine room which doesn't suck?
* Update the wiki.
* Sort through keyboards in the office
* Clean up wiki vandalism
* Clean up wiki vandalism
* Centralized repo for various configs: NFS, PAM auth with kerb, LDAP and Kerb5, routing/interfaces files
* Fix debian.csclub, aka our personal Debian repo which serves the CEO package
** Fix ceo versioning, which seems to be different on every machine it's installed on...
** Private subnet routing is broken on every machine ''except'' corn-syrup, taurine (see 'ethcrazy')

* Fix audio auth: audio is both a system group and an LDAP group and this has bad consequences for audio authorization
== Wiki Updates ==
* Centralized repo for various configs: NFS, PAM auth with kerb, /etc/hosts/, LDAP and Kerb5, routing/interfaces files

** LDAP login is currently broken on glomag, it is password with root only
* Update the following wiki pages
** Private subnet routing is broken on every machine ''except'' corn-syrup (see 'ethcrazy')
** [[Backups]]
* Update hosts list (10.15.134.WTF?)
** [[Ceo]], also related to "debian.csclub is broken"
** somehow merge [[Conserver]]/[[Serial Connections]]/[[Console Configuration]]
** [[Cscbot]]
** [[DNS]]
** [[Hardware]]
** [[Machine List]]
** [[Mirror]] including that rsyncd needs to be re-started after reboot. (systemctl start rsync)
** [[Music]] and possibly link to [[Pulseaudio]]
** [[MySQL]] mentions a-f replica
** [[NFS/Kerberos]] should probably by merged with an existing page
** [[NetApp]]
** [[Netboot]] (might want to merge with [[New CSC Machine]])
** [[OID Assignment]] and [[UID/GID Assignment]] should be merged with LDAP and replaced with a redirect
** Merge [[Point Of Sale]] and [[Point of Sale System]]
** [[Projects]]
** [[Scratch]]
** Add more info to [[Switches]]
** [[Virtualization]]
** [[Wireless]]
* Reorganize in general. Separate Machine/System Documentation section into multiple sections.
** Idea: One part for members/users, one part for office admin stuff, one part for Linux machines admin stuff, one part for CloudStack


==When in the Machine Room==
==When in the Machine Room==
* Make sure that the console connections are correct, up-to-date, and working.
* Set up binaerpilot.
* Decom aspartame
* Make sure that the IPMI/console connections are correct, up-to-date, and working.
* Label bays / server backs
* Fix psilodump's and aspartame's IPs and routing
* Take pictures of cables
** psilodump should not be routable outside aspartame. This is currently accomplished by fuckery. This *should* be fixed to use the net.ipv4.conf.all.arp_filter sysctl.
* Move alpha server to office
* Look into expanding /scratch and using RAID using spare disks in the office.
* Install new graphics card


==Science Machine Room==
==Science Machine Room==
* Replace cobalmin with better server
* Set up remote syslog2

Latest revision as of 16:35, 25 April 2023

These are things that syscom should do eventually:

Timeline

Get Mac Minis: Week of May 1

Get Intel Nucs: At least August but tentative

Get Rack Server: May

Get Netapp: Late May but tentative

Server Room Trip: May

General

  • Prepare for `sodium-benzoate` decom.
    • Need to find the physical server
    • Give to CSCF for decom
  • `/users` backups
  • Disaster recovery plan
  • Find out status of backup mirror
  • Get an IP/KVM for the machine room which doesn't suck?
  • Clean up wiki vandalism
  • Centralized repo for various configs: NFS, PAM auth with kerb, LDAP and Kerb5, routing/interfaces files
    • Private subnet routing is broken on every machine except corn-syrup, taurine (see 'ethcrazy')

Wiki Updates

When in the Machine Room

  • Make sure that the console connections are correct, up-to-date, and working.
  • Decom aspartame
  • Label bays / server backs
  • Take pictures of cables
  • Move alpha server to office
  • Install new graphics card

Science Machine Room

  • Replace cobalmin with better server