Hackeriet

User Tools

Site Tools

You are here: start » infra » operations » proxmox-maintenance
infra:operations:proxmox-maintenance

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
infra:operations:proxmox-maintenance [2026/05/31 17:05] – Document 2026-05-31 Proxmox maintenance run atluxity_idp.hackeriet.noinfra:operations:proxmox-maintenance [2026/06/01 17:32] (current) – Note LibreNMS alerting follow-up for klynge001 atluxity_idp.hackeriet.no
Line 1: Line 1:
 ====== Proxmox maintenance ====== ====== Proxmox maintenance ======
  
-This is a runbook for planned maintenance on the Hackeriet Proxmox hosts in [[infra:clusters:klynge001|klynge001]]. It is documentation and procedure, not inventory. Use NetBox for canonical device, IP, cabling, and VM placement data, but verify live Proxmox state before deciding maintenance impact.+This page documents maintenance for the Hackeriet Proxmox hosts in [[infra:clusters:klynge001|klynge001]]. It is a runbook and maintenance log.
  
 ===== Current scope ===== ===== Current scope =====
  
-Planned hosts:+Hosts currently covered by this procedure:
  
   * [[infra:hosts:host006|host006]]   * [[infra:hosts:host006|host006]]
   * [[infra:hosts:host007|host007]]   * [[infra:hosts:host007|host007]]
  
-Current goals: 
  
-  * Bring host006 and host007 up to date. +===== Last maintenance: 2026-05-31 =====
-  * Review failed services and storage health. +
-  * Keep the cluster healthy while working one host at a time. +
-  * Avoid guest-level changes unless needed for recovery.+
  
-This page covers Proxmox host maintenance. Service hosts that have moved away from this cluster should be checked as external service reachability, not treated as direct reboot impact from host006 or host007.+Scope: host006 and host007 in the ''klynge001'' Proxmox cluster.
  
-===== Announcement draft =====+Actions performed:
  
-Planned Proxmox maintenance for Hackeriet+  * Removed obsolete local backup archives on host006 for moved/stopped service VMs. 
 +  * Updated the scheduled Proxmox backup job to exclude moved/stopped VMs: 
 +    * ''105'' / ''blade'' 
 +    * ''510'' / ''ingress'' 
 +    * ''511'' / ''app-01'' 
 +  * Upgraded host006 and host007 to Proxmox VE 8.4.19. 
 +  * Rebooted host006 and host007 one at a time to activate kernel 6.8.12-28-pve. 
 +  * Temporarily adjusted expected votes during single-node reboot windows so the remaining node stayed quorate. 
 +  * Verified storage, package state, cluster quorum, guest state, and basic guest reachability after reboots.
  
-I plan to do maintenance on the Proxmox hosts host006 and host007 in the klynge001 cluster one of the next days.+Final state after maintenance:
  
-Scope:+  * host006: Proxmox VE 8.4.19, kernel 6.8.12-28-pve. 
 +  * host007: Proxmox VE 8.4.19, kernel 6.8.12-28-pve. 
 +  * Cluster: 2 nodes, expected votes 2, quorate. 
 +  * No pending package upgrades were listed on either node. 
 +  * host006 root filesystem usage was about 72% after cleanup and updates. 
 +  * host007 root filesystem usage was about 19% after updates.
  
-  * OS and Proxmox package updates +Follow-up actions completed after the maintenance:
-  * storage and backup health checks +
-  * failed service review +
-  * possible host reboots if required+
  
-Expected impact:+  * Retired Munin on host006 and host007 by disabling and stopping ''munin-node.service''. 
 +  * Cleared stale ''lock: migrate'' locks on moved/stopped VMs. 
 +  * Disabled autostart for moved/stopped VMs: 
 +    * ''105'' / ''blade'' 
 +    * ''510'' / ''ingress'' 
 +    * ''511'' / ''app-01'' 
 +    * ''601'' / ''idp1'' 
 +  * Verified that ''systemctl --failed'' was clear on both nodes after Munin retirement.
  
-  * VMs still running on host006 or host007 may be briefly unavailable if their host is rebooted. +===== What we learned =====
-  * VMs affected by a reboot or migration will be checked from Proxmox and, when possible, by guest-agent and network probes. +
-  * I will avoid guest-level changes unless needed for recovery. +
-  * I will work on one host at a time and check cluster health between steps.+
  
-===== Current service placement notes =====+  * The cluster is currently operating as a two-node cluster. During a single-node reboot, the remaining node can temporarily lose quorum unless expected votes is adjusted when either other node is rebooted. 
 +  * Expected votes returned to 2 after both nodes were back and joined. 
 +  * ''munin-node.service'' had been failing for months on both host006 and host007. It is now intentionally retired on these hosts. 
 +  * Moved/stopped service VMs may still carry stale Proxmox migration locks and ''onboot: 1'' from before migration. Clear stale locks only after confirming there is no active migration task. 
 +  * host007 emitted GRUB/LVM warnings about a missing physical volume name ''pv1'' during update-grub, but rebooted successfully on the new kernel. The active LVM metadata still contains an internal ''pv1'' label. 
 +  * host006 emitted a GRUB warning that the removable EFI fallback path is not updated automatically. The explicit Proxmox EFI boot entry worked and the host rebooted successfully.
  
-Observed on 2026-05-27:+===== Maintenance procedure =====
  
-  * ''ingress'' is stopped in Proxmox on host006, while [[infra:hosts:ingress|ingress.hackeriet.no]] is reachable elsewhere. +Work one host at a time. Do not reboot both host006 and host007 at once.
-  * ''app-01'' is stopped in Proxmox on host006, while [[infra:hosts:app-01|app-01.hackeriet.no]] is reachable elsewhere. +
-  * ''blade'' is stopped in Proxmox on host007, while blade.hackeriet.no and public services hosted there are reachable elsewhere. +
-  * ''idp1'' is stopped in Proxmox on host007, while idp.hackeriet.no is reachable. +
- +
-Current implication: +
- +
-  * Do not list moved services as expected reboot impact from host006 or host007. +
-  * Use live ''qm list'' output, not stale inventory records, to decide what a Proxmox reboot will affect. +
-  * Keep public service checks as regression checks, because DNS, routing, or documentation can still be wrong. +
- +
-NetBox still appears to contain stale VM placement for some moved systems. Treat NetBox as inventory to clean up later, not as the final maintenance-impact source for those systems. +
- +
-===== Current Proxmox impact ===== +
- +
-Before maintenance, refresh this with live ''qm list'' on each node. +
- +
-Observed on 2026-05-27, host006 running guests included: +
- +
-  * ''matrix'' +
-  * ''aleksei'' +
-  * ''login'' +
-  * ''jakkn'' +
-  * ''m'' +
-  * ''sub-planar'' +
-  * ''emma'' +
-  * ''foxboron'' +
-  * ''meshtastic-br'' +
- +
-Observed on 2026-05-27, host007 running guests included: +
- +
-  * ''jakkn2'' +
-  * ''fr30n'' +
-  * ''bergh'' +
-  * ''haus-forum'' +
-  * ''haus'' +
-  * ''hyperboria'' +
-  * ''nux'' +
-  * ''chat'' +
- +
-Stopped legacy or moved VMs should remain untouched unless the maintenance task explicitly includes cleanup or migration review. +
- +
-===== Affected guest checks ===== +
- +
-For host006/host007 maintenance, check guests that are actually expected to be affected by the Proxmox work. Do not use unrelated public service checks as the default success criteria. +
- +
-Use this order: +
- +
-  - Confirm the VM is expected to run with ''qm list''. +
-  - Confirm Proxmox sees the VM running with ''qm status <vmid>''. +
-  - If the QEMU guest agent is configured, check it with ''qm agent <vmid> ping''. +
-  - If the guest agent responds, discover guest IPs with ''qm guest cmd <vmid> network-get-interfaces''. +
-  - If the guest uses cloud-init IP config, check ''qm config <vmid>'' for ''ipconfig0''. +
-  - If needed, fall back to NetBox or neighbor/ARP information for the guest IP. +
-  - Probe network reachability only when it is expected for that guest. +
- +
-Useful commands: +
- +
-<code> +
-qm list +
-qm status <vmid> +
-qm agent <vmid> ping +
-qm guest cmd <vmid> network-get-interfaces +
-qm config <vmid> | sed -n '/^ipconfig/p;/^net/p' +
-ping -c 2 <ip> +
-nc -vz -w3 <ip> 22 +
-</code> +
- +
-Interpretation: +
- +
-  * ''qm status running'' means the hypervisor sees the VM running. +
-  * ''qm agent <vmid> ping'' means the guest OS and guest agent are responsive. +
-  * ''ping'' means basic network path works, if ICMP is allowed. +
-  * ''nc -vz -w3 <ip> 22'' means SSH is listening, if SSH is expected for that guest. +
-  * Lack of ping or SSH does not always mean the guest is broken; only investigate failures that contradict expectations for that guest. +
- +
-Observed on 2026-05-31, guest agent responded for: +
- +
-  * ''matrix'' +
-  * ''login'' +
-  * ''jakkn'' +
-  * ''m'' +
-  * ''sub-planar'' +
-  * ''emma'' +
-  * ''foxboron'' +
-  * ''meshtastic-br'' +
-  * ''jakkn2'' +
-  * ''bergh'' +
-  * ''haus-forum'' +
-  * ''hyperboria'' +
- +
-Observed on 2026-05-31, guest agent did not respond for: +
- +
-  * ''aleksei'' +
-  * ''fr30n'' +
-  * ''haus'' +
-  * ''nux'' +
-  * ''chat'' +
- +
-DNS checks are only relevant if DNS, routing, authoritative nameservers, or hosts serving those services are in scope. They are not default host006/host007 Proxmox update checks. +
- +
-===== Pre-maintenance checks =====+
  
-Run on both host006 and host007 before making changes:+Before making changes on either host:
  
 <code> <code>
 hostname -f hostname -f
 pveversion -v pveversion -v
 +uname -r
 pvecm status pvecm status
 systemctl --failed --no-pager systemctl --failed --no-pager
 pvesm status pvesm status
-df -h+df -h -x tmpfs -x devtmpfs
 qm list qm list
 cat /etc/pve/jobs.cfg cat /etc/pve/jobs.cfg
Line 162: Line 77:
 </code> </code>
  
-On host006, also check local storage pressure:+For host006, also check local storage pressure:
  
 <code> <code>
Line 168: Line 83:
 </code> </code>
  
-Host006 storage and backup context is documented at [[infra:operations:proxmox-backups|Proxmox backups]].+Update flow for each host:
  
-===== Maintenance procedure ===== +  - Confirm cluster health with ''pvecm status''. 
- +  - Confirm storage health with ''pvesm status'' and ''df -h''.
-Work one host at a time. Do not reboot both host006 and host007 at once. +
- +
-Suggested order: +
- +
-  - Start with host006 if the main concern is storage and backup health. +
-  - Start with host007 if its running guests need to be handled first. +
- +
-For each host: +
- +
-  - Confirm cluster state with ''pvecm status''. +
-  - Confirm storage state with ''pvesm status'' and ''df -h''.+
   - Review running guests with ''qm list''.   - Review running guests with ''qm list''.
-  - Confirm no unexpected critical service VM is running on the host before rebooting. 
   - Review failed units with ''systemctl --failed --no-pager''.   - Review failed units with ''systemctl --failed --no-pager''.
-  - Run ''apt update''. +  - Simulate package changes if the update set is large or risky.
-  - Review ''apt list --upgradable''.+
   - Apply updates only after reviewing the package set.   - Apply updates only after reviewing the package set.
-  - Reboot only if required or clearly useful.+  - Reboot only if required or clearly useful, such as after a kernel update.
   - After reboot, wait for the node to return and confirm cluster health before touching the next host.   - After reboot, wait for the node to return and confirm cluster health before touching the next host.
  
-Suggested update command, after review:+Suggested commands after review:
  
 <code> <code>
 +apt-get -s full-upgrade
 apt full-upgrade apt full-upgrade
 </code> </code>
  
-Do not change guest VM configuration as part of host maintenance unless needed for recovery.+===== Quorum during reboot ===== 
 + 
 +When one node is rebooted, the remaining node may temporarily lose quorum. If that happens during planned maintenance, set expected votes to 1 on the remaining node: 
 + 
 +<code> 
 +pvecm expected 1 
 +pvecm status 
 +</code> 
 + 
 +After the rebooted node rejoins, confirm the cluster has returned to two nodes and expected votes 2: 
 + 
 +<code> 
 +pvecm status 
 +</code> 
 + 
 +Do not use this as an incident workaround without understanding which node has the correct cluster state.
  
 ===== Post-host checks ===== ===== Post-host checks =====
Line 206: Line 124:
 <code> <code>
 hostname -f hostname -f
-pveversion -v+pveversion 
 +uname -r
 pvecm status pvecm status
 systemctl --failed --no-pager systemctl --failed --no-pager
 pvesm status pvesm status
-df -h+df -h -x tmpfs -x devtmpfs
 qm list qm list
 +apt list --upgradable
 </code> </code>
  
Line 217: Line 137:
  
 <code> <code>
-qm list 
 qm status <vmid> qm status <vmid>
 qm agent <vmid> ping qm agent <vmid> ping
Line 226: Line 145:
 </code> </code>
  
-For migrated VMs, check both source and target:+Interpretation:
  
-<code> +  * ''qm status running'' means the hypervisor sees the VM running. 
-qm list    # source: VM should be stopped or absent from the source host +  * ''qm agent <vmid> ping'' means the guest OS and QEMU guest agent are responsive. 
-qm list    # target: VM should be running on the target host +  * ''ping'' means basic network path works, if ICMP is allowed. 
-</code>+  * ''nc -vz -w3 <ip> 22'' means SSH is listening, if SSH is expected for that guest. 
 +  * Lack of ping or SSH does not always mean the guest is broken. Investigate only failures that contradict expectations for that guest.
  
-Check guest services only when the relevant guest has been touched and the expected service is known.+===== Monitoring / LibreNMS =====
  
-===== Known caveats =====+As of 2026-06-01, host006 and host007 are monitored in LibreNMS as Proxmox hypervisors. This replaces the old Munin host monitoring for these nodes.
  
-  * host006 has about 1 TB physical storage, but Proxmox local storage is on the root filesystem. The root filesystem was previously close to full, and local backups under ''/var/lib/vz/dump'' were the main pressure point. +LibreNMS records:
-  * During host006 maintenance, avoid casual LVM reshaping. It can put VM disks at risk and should only be done with a maintenance window and recovery plan. +
-  * Certificate automation for internal Proxmox hostnames is documented at [[infra:operations:proxmox-acme-dns|Proxmox ACME DNS automation]]. +
-  * NetBox placement may be stale for moved service VMs. Verify live Proxmox state with ''qm list''.+
  
-===== Safety notes =====+  * host006.hackeriet.no: LibreNMS device ID 25. 
 +  * host007.hackeriet.no: LibreNMS device ID 26. 
 +  * Both devices have OS detected as ''proxmox''. 
 +  * Both devices have the LibreNMS ''proxmox'' application enabled with app instance ''klynge001''.
  
-  * There is no plan, and we should avoid, to touch guest VMs unless required for recovery. +Host-side setup:
-  * Do not change DNS during the maintenance window unless DNS itself is the incident. +
-  * There is no plan, and we should avoid, delete backups or ISOs without understanding what they are. +
-  * Keep notes locally while working; wiki, NetBox, IDP, and public services may be affected depending on which host is down.+
  
-===== Maintenance log: 2026-05-31 =====+  * ''snmpd'' is installed, enabled, and running on both hosts. 
 +  * SNMP listens only on the management IP of each host: 
 +    * host006: ''10.10.50.26:161/udp'' 
 +    * host007: ''10.10.50.27:161/udp'' 
 +  * SNMP uses SNMPv3 authPriv with the security name ''librenms_klynge001''. 
 +  * Credentials are stored in Hackerpass at ''infrastructure/librenms-klynge001-snmpv3''. Do not put the credential values in the wiki or NetBox. 
 +  * The Proxmox LibreNMS agent-local script is installed as ''/usr/local/libexec/librenms-proxmox''. 
 +  * SNMP exposes Proxmox VM traffic through:
  
-Scope: host006 and host007 in the klynge001 Proxmox cluster.+<code> 
 +extend proxmox /usr/bin/sudo /usr/local/libexec/librenms-proxmox 
 +</code>
  
-Actions performed:+  * ''/etc/sudoers.d/librenms-proxmox'' allows the ''Debian-snmp'' user to run only that script via sudo.
  
-  * Removed obsolete local backup archives on host006 for moved/stopped service VMs. +Firewall setup:
-  * Updated the scheduled Proxmox backup job to exclude moved/stopped VMs: +
-    * 105 / blade +
-    * 510 / ingress +
-    * 511 / app-01 +
-  * Upgraded host006 and host007 to Proxmox VE 8.4.19. +
-  * Rebooted host006 and host007 one at a time to activate kernel 6.8.12-28-pve. +
-  * Temporarily adjusted expected votes during single-node reboot windows so the remaining node stayed quorate. +
-  * Verified storage, package state, cluster quorum, guest state, and basic guest reachability after reboots.+
  
-Final state after the maintenance:+  * The Proxmox cluster firewall allows SNMP only from app-01 / LibreNMS source address ''10.10.50.51'' to the ''klynge001'' IP set. 
 +  * The firewall rule is in ''/etc/pve/firewall/cluster.fw'':
  
-  * host006: Proxmox VE 8.4.19, running kernel 6.8.12-28-pve. +<code> 
-  * host007: Proxmox VE 8.4.19, running kernel 6.8.12-28-pve. +IN ACCEPT -source 10.10.50.51/32 -dest +klynge001 -p udp -dport 161 -log nolog # LibreNMS SNMP polling from app-01 
-  * Cluster: 2 nodes, expected votes 2, quorate. +</code>
-  * No pending package upgrades were listed on either node after maintenance. +
-  * Expected running guests were back online after the host reboots. +
-  * host006 root filesystem pressure was reduced from the previous critical state to about 72% used after cleanup and updates. +
-  * host007 root filesystem was about 19% used after updates.+
  
-Guest checks used:+LibreNMS setup:
  
-  * qm list to confirm expected running/stopped VM state. +  * ''enable_proxmox'' is set to ''true''. 
-  * qm agent <vmid> ping where guest agent was available. +  * The ''unix-agent'' poller module is disabled per device for host006 and host007, because this setup uses SNMP extend instead of the LibreNMS unix-agent on port 6556. 
-  * qm guest cmd <vmid> network-get-interfaces where useful for IP discovery. +  * No LibreNMS alert rules existed when this monitoring was added. Host metrics and Proxmox application polling are live, but alert policy still needs to be defined.
-  * ping and nc -z -w2 <ip> 22 for lightweight reachability checks where those probes matched expectations.+
  
-Operational notes learned:+Useful verification commands:
  
-  * The cluster is currently operating as a two-node cluster. During planned single-node reboot windows, the remaining node can temporarily lose quorum unless expected votes is adjusted. +<code> 
-  * Expected votes returned to 2 automatically after both nodes were back and joined. +# On each Proxmox host 
-  * Some guests did not immediately appear as running right after host reboot, but came up shortly afterward. Wait briefly before manual intervention. +systemctl is-active snmpd 
-  * Lack of ICMP is not by itself a failure. For example, haus-forum did not answer ping after reboot but SSH was open. +ss -lunp | grep ':161' 
-  * munin-node.service is failed on both host006 and host007. This was pre-existing or at least unrelated to the package upgrade and remains unresolved. +sudo -u Debian-snmp sudo /usr/local/libexec/librenms-proxmox
-  * host007 emitted GRUB/LVM warnings about a missing physical volume name pv1 during update-grub, but rebooted successfully on the new kernel. +
-  * host006 emitted a GRUB warning that the removable EFI fallback path is not updated automatically. The explicit Proxmox EFI boot entry worked and the host rebooted successfully.+
  
-Follow-up items:+# From the LibreNMS container on app-01 
 +snmpwalk -v3 -l authPriv -u librenms_klynge001 -a SHA -A '<auth password>' -x AES -X '<privacy password>' host006.hackeriet.no SNMPv2-MIB::sysName.0 
 +snmpget -v3 -l authPriv -u librenms_klynge001 -a SHA -A '<auth password>' -x AES -X '<privacy password>' -Oqv host006.hackeriet.no .1.3.6.1.4.1.8072.1.3.2.3.1.2.7.112.114.111.120.109.111.120 
 +lnms device:poll -m applications host006.hackeriet.no 
 +</code>
  
-  * Fix or intentionally retire Munin on host006 and host007. +===== Follow-up items ===== 
-  * Review VM autostart settings for guests expected to return after host reboot. + 
-  * Investigate the host007 GRUB/LVM pv1 warning.+  * Define LibreNMS alert rules and notification routing for klynge001 host monitoring. 
 +  * Investigate the host007 GRUB/LVM ''pv1'' warning.
   * Investigate the host006 EFI fallback warning.   * Investigate the host006 EFI fallback warning.
-  * Document the expected-votes procedure explicitly if this remains a two-node maintenance pattern.+  * Review whether the two-node quorum procedure should be part of the standard klynge001 maintenance runbook. 
 +  * Keep NetBox VM placement aligned with the current reality for moved service VMs. 
 +  * If Munin package cleanup is desired later, remove it in a separate low-risk cleanup task. 
 + 
 +===== Other notes ===== 
 + 
 +  * Host006 storage and backup context is documented at [[infra:operations:proxmox-backups|Proxmox backups]]. 
 +  * Certificate automation for internal Proxmox hostnames is documented at [[infra:operations:proxmox-acme-dns|Proxmox ACME DNS automation]].
  
/srv/hackeriet-wiki/dokuwiki/data/attic/infra/operations/proxmox-maintenance.1780247145.txt.gz · Last modified: by atluxity_idp.hackeriet.no