Hello,
I’m trying to speak from a Fly.io VM in singapore to a vpn endpoint in Frankfurt over your wireguard network. The speed is abysmal, 20-40 KBps.
10MB.dd 39%[=====> ] 3.91M 34.1KB/s eta 3m 57s
When moving the Fly.io VM to Toronto and keeping the same Frankfurt vpn endpoint everything is fine:
2022-01-23 13:03:49 (7.56 MB/s) - '10MB.dd' saved [10485760/10485760]
Pings are normal.
Is this to be expected?
Hi,
Assuming you created this peer in the last 24h or so, is it an option to re-create the wg peer from sin
to fra
?
Although the comment below talks about wg peering, it mentions only ams
, so I don’t know if it’s applicable to fra
.
You should be OK again. You’re catching the tail end of a migration we’re doing to a new system for managing WireGuard peers (as of tomorrow, it should take single-digit seconds to create and bring up WireGuard for a new peer, rather than 1.5 minutes).
I think you created and then deleted a peer, which snagged a corner case — I had a UNIQUE constraint on a sqlite column that the peer you’re talking about was hitting.
I’ve rolled back AMS to the old (slow) behavior (boo!) — it should be fine fo…
1 Like
The fra
peer is ~1 week old. No problem deleting and recreating it, but I don’t remember it taking 1.5 minutes to do so. I think maybe this is unrelated?
kurt
January 24, 2022, 6:27pm
4
This seems like a problem. Can you explain the exact setup you’re using to test this? What’s running on Fly, how are you connecting to it, where does the VPN endpoint connect?
cedivad
January 24, 2022, 10:57pm
5
Now that everyone in Asia is asleep things are fine.
This shows it takes 3x as long for a http get request during busier hours:
Maybe I will just look for a better connected region