|
Bogdan Timofte
authored
3 months ago
|
1
|
# pve-net-hang-watchdog
|
|
|
2
|
|
|
|
3
|
`pve-net-hang-watchdog` este un serviciu simplu care urmareste jurnalul kernel pentru hang-uri de NIC si incearca recuperarea uplink-ului prin `ifdown` si `ifup`.
|
|
|
4
|
|
|
|
5
|
## Rol
|
|
|
6
|
|
|
|
7
|
Util pentru noduri Proxmox unde interfata fizica din spatele unui bridge WAN poate intra in stare de hang hardware, iar recovery-ul cel mai pragmatic este ciclarea link-ului.
|
|
|
8
|
|
|
|
9
|
## Componente
|
|
|
10
|
|
|
|
11
|
- [bin/pve-net-hang-watchdog.sh](/Users/bogdan/Documents/Workspaces/Xdev/Madagascar/cluster/projects/pve-net-hang-watchdog/bin/pve-net-hang-watchdog.sh) - scriptul principal
|
|
|
12
|
- [systemd/pve-net-hang-watchdog.service](/Users/bogdan/Documents/Workspaces/Xdev/Madagascar/cluster/projects/pve-net-hang-watchdog/systemd/pve-net-hang-watchdog.service) - unitatea systemd
|
|
|
13
|
- [config/xdev-pve-net-hang-watchdog](/Users/bogdan/Documents/Workspaces/Xdev/Madagascar/cluster/projects/pve-net-hang-watchdog/config/xdev-pve-net-hang-watchdog) - defaults standard
|
|
|
14
|
- [scripts/install.sh](/Users/bogdan/Documents/Workspaces/Xdev/Madagascar/cluster/projects/pve-net-hang-watchdog/scripts/install.sh) - install local
|
|
|
15
|
- [scripts/uninstall.sh](/Users/bogdan/Documents/Workspaces/Xdev/Madagascar/cluster/projects/pve-net-hang-watchdog/scripts/uninstall.sh) - uninstall canonic
|
|
|
16
|
- [setup.sh](/Users/bogdan/Documents/Workspaces/Xdev/Madagascar/cluster/projects/pve-net-hang-watchdog/setup.sh) - wrapper local/remote
|
|
|
17
|
|
|
|
18
|
## Locatii instalate pe host
|
|
|
19
|
|
|
|
20
|
- comanda/daemon script: `/usr/local/sbin/pve-net-hang-watchdog.sh`
|
|
|
21
|
- uninstall canonic: `/usr/local/lib/xdev/pve-net-hang-watchdog/uninstall.sh`
|
|
|
22
|
- wrapper optional pentru uninstall: `/usr/local/sbin/xdev-pve-net-hang-watchdog-uninstall`
|
|
|
23
|
- defaults: `/etc/default/xdev-pve-net-hang-watchdog`
|
|
|
24
|
- unitate systemd: `/etc/systemd/system/pve-net-hang-watchdog.service`
|
|
|
25
|
- documentatie instalata: `/usr/local/share/doc/xdev/pve-net-hang-watchdog`
|
|
|
26
|
|
|
|
27
|
## Configurare
|
|
|
28
|
|
|
|
29
|
Parametri suportati prin defaults:
|
|
|
30
|
|
|
|
31
|
- `WATCH_BRIDGE`
|
|
|
32
|
- `WATCH_IFACE`
|
|
|
33
|
- `COOLDOWN_SECONDS`
|
|
|
34
|
- `HANG_PATTERN`
|
|
|
35
|
|
|
|
36
|
Daca `WATCH_IFACE` este gol, scriptul incearca sa descopere automat interfata fizica din `bridge-ports`.
|
|
|
37
|
|
|
|
38
|
## Flux rapid
|
|
|
39
|
|
|
|
40
|
```bash
|
|
|
41
|
sudo ./setup.sh --local
|
|
|
42
|
sudo systemctl status pve-net-hang-watchdog.service
|
|
|
43
|
```
|
|
|
44
|
|
|
|
45
|
## Operare
|
|
|
46
|
|
|
|
47
|
Loguri:
|
|
|
48
|
|
|
|
49
|
```bash
|
|
|
50
|
journalctl -u pve-net-hang-watchdog.service -f
|
|
|
51
|
```
|
|
|
52
|
|
|
|
53
|
Configurare:
|
|
|
54
|
|
|
|
55
|
```bash
|
|
|
56
|
sudo editor /etc/default/xdev-pve-net-hang-watchdog
|
|
|
57
|
sudo systemctl restart pve-net-hang-watchdog.service
|
|
|
58
|
```
|
|
|
59
|
|
|
|
60
|
Installerul face si `enable --now`, deci dupa instalare serviciul este deja pornit.
|
|
|
61
|
|
|
|
62
|
## Uninstall
|
|
|
63
|
|
|
|
64
|
```bash
|
|
|
65
|
sudo ./setup.sh --local --uninstall
|
|
|
66
|
```
|
|
|
67
|
|
|
|
68
|
Sau direct:
|
|
|
69
|
|
|
|
70
|
```bash
|
|
|
71
|
sudo /usr/local/lib/xdev/pve-net-hang-watchdog/uninstall.sh
|
|
|
72
|
```
|