Qnap iSCSI mag keine Jumbo Frames

CoMo

Commander
Registriert
Dez. 2015
Beiträge
3.006
Hallo,

ich habe hier ein altes Qnap TS-419P II NAS. Eine Festplatte davon reiche ich als iSCSI-Gerät an Proxmox durch.

Sobald ich die MTU auf dem NAS höher als 1500 stelle, kommt keine iSCSI-Verbindung mehr zustande. Proxmox sagt:

Code:
Feb 20 15:08:19 pve iscsid[17797]: Kernel reported iSCSI connection 1:0 error (1022 - ISCSI_ERR_NOP_TIMEDOUT: A NOP has timed out) state (3)
Feb 20 15:08:22 pve pveproxy[1466]: proxy detected vanished client connection
Feb 20 15:08:27 pve iscsid[17797]: connection1:0 is operational after recovery (1 attempts)
Feb 20 15:08:37 pve kernel:  connection1:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4297304399, last ping 4297309696, now 4297314816
Feb 20 15:08:37 pve kernel:  connection1:0: detected conn error (1022)
Feb 20 15:08:38 pve iscsid[17797]: Kernel reported iSCSI connection 1:0 error (1022 - ISCSI_ERR_NOP_TIMEDOUT: A NOP has timed out) state (3)
Feb 20 15:08:40 pve iscsid[17797]: connection1:0 is operational after recovery (1 attempts)
Feb 20 15:08:49 pve kernel:  connection1:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4297316827, last ping 4297321984, now 4297327104
Feb 20 15:08:49 pve kernel:  connection1:0: detected conn error (1022)
Feb 20 15:08:49 pve kernel: sd 0:0:0:0: [sda] tag#39 FAILED Result: hostbyte=DID_TRANSPORT_DISRUPTED driverbyte=DRIVER_OK cmd_age=77s
Feb 20 15:08:49 pve kernel: sd 0:0:0:0: [sda] tag#39 CDB: Read(10) 28 00 00 00 00 00 00 01 00 00
Feb 20 15:08:49 pve kernel: I/O error, dev sda, sector 0 op 0x0:(READ) flags 0x0 phys_seg 32 prio class 0
Feb 20 15:08:49 pve kernel: sd 0:0:0:0: [sda] tag#40 FAILED Result: hostbyte=DID_TRANSPORT_DISRUPTED driverbyte=DRIVER_OK cmd_age=77s
Feb 20 15:08:49 pve kernel: sd 0:0:0:0: [sda] tag#40 CDB: Read(10) 28 00 00 00 00 00 00 01 00 00
Feb 20 15:08:49 pve kernel: I/O error, dev sda, sector 0 op 0x0:(READ) flags 0x0 phys_seg 32 prio class 0
Feb 20 15:08:50 pve iscsid[17797]: Kernel reported iSCSI connection 1:0 error (1022 - ISCSI_ERR_NOP_TIMEDOUT: A NOP has timed out) state (3)
Feb 20 15:08:52 pve iscsid[17797]: connection1:0 is operational after recovery (1 attempts)
Feb 20 15:09:01 pve kernel:  connection1:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4297329107, last ping 4297334272, now 4297339392
Feb 20 15:09:01 pve kernel:  connection1:0: detected conn error (1022)
Feb 20 15:09:02 pve iscsid[17797]: Kernel reported iSCSI connection 1:0 error (1022 - ISCSI_ERR_NOP_TIMEDOUT: A NOP has timed out) state (3)
Feb 20 15:09:04 pve iscsid[17797]: connection1:0 is operational after recovery (1 attempts)
Feb 20 15:09:14 pve kernel:  connection1:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4297341403, last ping 4297346560, now 4297351680
Feb 20 15:09:14 pve kernel:  connection1:0: detected conn error (1022)

Kennt jemand dafür eine Lösung? Muss ich in Proxmox irgendwas konfigurieren?
 
Alle Komponenten auf der Strecke (also auch Switche, Router, etc.) müssen mit Jumobframes umgehen können und auf einen gemenisamen Nenner konfiguriert sein. Hast du die Einstellung auf dem PVE auch passend gesetzt? Nutzt du bereits erfolgreich mit anderen Komponenten Jumboframes?
 
Zuletzt bearbeitet: (Konkretisiert, was "alle Komponenten" sind.)
Dazu fallen mir 2 Themengebiete ein:

1. Läuft die Verbindung über eine Fritzbox ?
Wenn ja, dann hat der TO leider Pech gehabt, da FB keine Jumbo Frames beherrscht.

2. NAS mit aktuellem QTS verwenden.
 
Zuletzt bearbeitet:
Ich kann meinen Router pingen:
Code:
ping -f -l 8972 opnsense
Ping wird ausgeführt für opnsense.home.arpa [192.168.100.1] mit 8972 Bytes Daten:
Antwort von 192.168.100.1: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.1: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.1: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.1: Bytes=8972 Zeit<1ms TTL=64

Meine Proxmox-Nodes:
Code:
ping -f -l 8972 pve
Ping wird ausgeführt für pve.home.arpa [192.168.100.100] mit 8972 Bytes Daten:
Antwort von 192.168.100.100: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.100: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.100: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.100: Bytes=8972 Zeit<1ms TTL=64

Code:
ping -f -l 8972 pve2
Ping wird ausgeführt für pve2.home.arpa [192.168.100.101] mit 8972 Bytes Daten:
Antwort von 192.168.100.101: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.101: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.101: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.101: Bytes=8972 Zeit<1ms TTL=64

LXC-Container auf dem Proxmox:


Code:
ping -f -l 8972 adguard
Ping wird ausgeführt für adguard.home.arpa [192.168.100.53] mit 8972 Bytes Daten:
Antwort von 192.168.100.53: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.53: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.53: Bytes=8972 Zeit<1ms TTL=64
Antwort von 192.168.100.53: Bytes=8972 Zeit<1ms TTL=64

Vom Router zu meinem PC:


Code:
root@OPNsense:~ # ping -D -s 8972 192.168.100.65
PING 192.168.100.65 (192.168.100.65): 8972 data bytes
8980 bytes from 192.168.100.65: icmp_seq=0 ttl=128 time=0.637 ms
8980 bytes from 192.168.100.65: icmp_seq=1 ttl=128 time=0.619 ms
8980 bytes from 192.168.100.65: icmp_seq=2 ttl=128 time=0.867 ms
8980 bytes from 192.168.100.65: icmp_seq=3 ttl=128 time=0.771 ms

Und auch das NAS frisst die MTU 9000. Also wird es wohl funktionieren.

Auch das NAS ist weiterhin erreichbar. Nur iSCSI funktioniert nicht mehr.

Alle Switches auf dem Weg unterstützen Jumbo Frames. Das NAS hängt mit 2 Gigabit Links und LACP am Switch, der Proxmox mit 10G DAC.

Wenn ich den Link auf dem NAS auf 1500 stelle, funktioniert es wieder. Dann sind die Pakete halt fragmentiert.
 
Zuletzt bearbeitet:
Zurück
Oben