Repository navigation
Closed
Labels
storageRelated to storage.Related to storage.
Description
Rack: rack2 (Dogfood)
While running sanity tests on Dogfood, we got an instance stuck in 'Creating'.
Upon checking further, looks like the saga got stuck while creating a region:
= INSTANCE ====================================================================
ID: 8719ab23-4949-4841-89f4-99e4401a7415
project ID: 290be4ae-3f36-40b0-9fc7-6d49277c9bf4
name: alldisks-instance-main-862916
description: test-instance
created at: 2026-09-22 18:54:16.575461 UTC
last modified at: 2026-09-22 18:54:16.575461 UTC
== CONFIGURATION ===============================================================
vCPUs: 4
memory: 16 GiB
hostname: test-host
boot disk: None
auto-restart:
InstanceAutoRestart {
policy: None,
cooldown: None,
}
== RUNTIME STATE ===============================================================
nexus state: Creating
(i) external API state: Creating
intended state: running
last updated at: 2026-09-22T18:54:16.575461Z (generation 1)
needs reincarnation: false
karmic status: saṃsāra (reincarnation enabled)
last reincarnated at: None
active VMM ID: None
target VMM ID: None
migration ID: None
updater lock: UNLOCKED at generation: 1
18:54:17.011Z DEBG ba00935f-ad63-4bc2-a61e-f459809f986e (ServerContext): authorize result
action = CreateChild
actor = Some(Actor::SiloUser { silo_user_id: 9ceeb971-95af-4cbe-8b25-0c9ad57c925b (silo_user), silo_id: 7bd7623a-68ed-4636-8ecb-b59e3b068787, .. })
authenticated = true
resource = Instance { parent: Project { parent: Silo { parent: Fleet, key: 7bd7623a-68ed-4636-8ecb-b59e3b068787, lookup_type: ById(7bd7623a-68ed-4636-8ecb-b59e3b068787) }, key: 290be4ae-3f36-40b0-9fc7-6d49277c9bf4, lookup_type: ById(290be4ae-3f36-40b0-9fc7-6d49277c9bf4) }, key: 8719ab23-4949-4841-89f4-99e4401a7415, lookup_type: ById(8719ab23-4949-4841-89f4-99e4401a7415) }
result = Ok(())
saga_id = 17156bc3-8b9a-47b0-bcee-a41ceca46211
saga_name = instance-create
saga_node = CreateNetworkInterface0
silo_id = 7bd7623a-68ed-4636-8ecb-b59e3b068787
silo_user_id = 9ceeb971-95af-4cbe-8b25-0c9ad57c925b
type = silo_user
18:56:19.123Z WARN ba00935f-ad63-4bc2-a61e-f459809f986e (ServerContext): failed to create region 7117425b-7aaa-4696-a700-b10665b52f7f, retrying in 118.445287179s
error = Communication Error: error sending request for url (http://[fd00:1122:3344:10b::7]:32345/crucible/0/regions): error sending request for url (http://[fd00:1122:3344:10b::7]:32345/crucible/0/regions): client error (Connect): tcp connect error: Connection refused (os error 146)
file = nexus/src/app/crucible.rs:209
saga_id = 17156bc3-8b9a-47b0-bcee-a41ceca46211
saga_name = instance-create
18:58:01.845Z INFO ba00935f-ad63-4bc2-a61e-f459809f986e (ServerContext): observed saga start
background_task = saga_recovery
file = nexus/saga-recovery/src/recovery.rs:78
saga_id = 17156bc3-8b9a-47b0-bcee-a41ceca46211
18:58:01.846Z DEBG ba00935f-ad63-4bc2-a61e-f459809f986e (ServerContext): found saga that can be ignored (already running)
background_task = saga_recovery
saga_id = 17156bc3-8b9a-47b0-bcee-a41ceca46211
18:58:17.583Z WARN ba00935f-ad63-4bc2-a61e-f459809f986e (ServerContext): failed to create region 7117425b-7aaa-4696-a700-b10665b52f7f, retrying in 283.290336713s
error = Communication Error: error sending request for url (http://[fd00:1122:3344:10b::7]:32345/crucible/0/regions): error sending request for url (http://[fd00:1122:3344:10b::7]:32345/crucible/0/regions): client error (Connect): tcp connect error: Connection refused (os error 146)
file = nexus/src/app/crucible.rs:209
saga_id = 17156bc3-8b9a-47b0-bcee-a41ceca46211
saga_name = instance-create
Metadata
Metadata
Assignees
Labels
storageRelated to storage.Related to storage.
Activity
paudmir commentedon Sep 22, 2026
hawkw commentedon Sep 22, 2026
It appears the saga is still running --- is it retrying the connection attempt indefinitely?
paudmir commentedon Sep 23, 2026
Looked like that, eventually the saga showed things wrapped up, there's just this 16 h gap in the disk_create.regions_ensure (which eventually failed).
hawkw commentedon Sep 23, 2026
wow! that's a while!
paudmir commentedon Sep 28, 2026
I'll close because it did get resolved but the '16 h waiting period' is still a mystery.
askfongjojo commentedon Oct 7, 2026
So the connection-refusing crucible-agent (http://[fd00:1122:3344:10b::7]:32345) was indeed the problem one on sled 14 that crashed repeatedly because the root dataset ran out of space. The stuck instance creation timestamps fall into the duration when crucible-agent stopped responding:
Crucible-agent recovered after I cleaned up the thousands of core files on that sled the next day.