Open tgodzik opened 4 years ago
Do we have logs for this error? Metals always “retries” by nature of calling coursier on every server start. I’m not sure we should retry on the metals side
What logs should I produce?
Do we have logs for this error? Metals always “retries” by nature of calling coursier on every server start. I’m not sure we should retry on the metals side
@olafurpg We need to run restart manually, it would be useful create a MessageRequest that provides us with a capability to restart but without relaoding
What logs should I produce?
@soronpo from .metals/metals.log
where the download failed
I got the part where it failed to download the Semantic DB:
The server is listening for incoming connections at tcp://127.0.0.1:34252...
Starting thread that pumps stdin and redirects it to the bsp server...
Starting thread that pumps server stdout and redirects it to the client stdout...
INFO tracing is disabled for protocol BSP, to enable tracing of incoming and outgoing JSON messages create an empty file at /home/soronpo/.cache/metals/bsp.trace.json
WARN Stopped configuration of SemanticDB in Scala 2.13.1 projects: Error downloading org.scalameta:semanticdb-scalac_2.13.1:4.3.0
not found: /home/soronpo/.ivy2/local/org.scalameta/semanticdb-scalac_2.13.1/4.3.0/ivys/ivy.xml
download error: Caught javax.net.ssl.SSLException: readHandshakeRecord (readHandshakeRecord) while downloading https://repo1.maven.org/maven2/org/scalameta/semanticdb-scalac_2.13.1/4.3.0/semanticdb-scalac_2.13.1-4.3.0.pom
not found: https://dl.bintray.com/scalacenter/releases/org/scalameta/semanticdb-scalac_2.13.1/4.3.0/semanticdb-scalac_2.13.1-4.3.0.pom
INFO time: connected to build server in 12s
@soronpo are you behind a VPN?
Possibly. This error is on my university's computer.
Similar issue here. Probably not metals' problem, rather coursier's issue but still. The downloads from repo1.maven.org
are known to connection-reset at times, and since install pulls a lot of jars at startup, it is not uncommon for server install to fail. On a bad day it happens to me 2-3 times in a row easily.
I do development in containers, so "Rebuild container" command typically helps, as mentioned, on 2nd-3rd try the worst case.
Perhaps joining forces with sbt (sbt-7200) might persuade the coursier implement transparent retries throughout?
Describe the bug
Downloading Metals might timeout and it's rather hard for user to retry.
Expected behavior
Message to the user with actionable option to retry downloading.
Additional context
Reported by @soronpo when updating to
0.8.0
Search terms
conenction internet download coursier