mirrors/scylladb
1
0
Fork 0
mirror of https://github.com/scylladb/scylladb.git synced 2026-05-24 00:32:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

main: better error message on failing to start Prometheus

Browse Source 
Previously, if the Prometheus port (by default, 0.0.0.0:9180) could not
be opened, the following message appeared in the log about 10 seconds into
the run, and Scylla crashed.

ERROR 2017-01-01 19:31:04,066 [shard 0] seastar - Exiting on unhandled exception: std::system_error (error system:98, Address already in use)

The puzzled user would have no idea *which* address was already in use, why,
or why Scylla stopped.

In this patch, before the above message we get the much more informative
message:

ERROR 2017-01-01 19:58:19,080 [shard 0] init - Could not start Prometheus API server on 0.0.0.0:9180: std::system_error (error system:98, Address already in use)

We continue to print the original message - and exit - in this case,
under the assumption that it's better not to run the database while
improperly configured.

Signed-off-by: Nadav Har'El <nyh@scylladb.com>
Message-Id: <20170102121304.2060-1-nyh@scylladb.com>
This commit is contained in:
Nadav Har'El
2017-01-02 14:13:04 +02:00
committed by Avi Kivity
parent 0c746b22e0
commit 45f19f2633
1 changed files with 4 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
main.cc
View File

@@ -689,7 +689,10 @@ int main(int ac, char** av) {
pctx.prefix = cfg->prometheus_prefix();
prometheus_server.start().get();
prometheus::start(prometheus_server, pctx);
prometheus_server.listen(ipv4_addr{prom_addr.addresses[0].in.s_addr, pport}).get();
prometheus_server.listen(ipv4_addr{prom_addr.addresses[0].in.s_addr, pport}).handle_exception([pport, &cfg] (auto ep) {
startlog.error("Could not start Prometheus API server on {}:{}: {}", cfg->prometheus_address(), pport, ep);
return make_exception_future<>(ep);
}).get();
}
supervisor_notify("serving");
// Register at_exit last, so that storage_service::drain_on_shutdown will be called first