lab/chart/ratelimit-test.sh, after frontends-test.sh:
The webmail's own limit counts per replica: replica A refuses one client at attempt 201, and replica B still starts its sign-in.
The webmail counts the browser its proxy names, not the proxy: replica A still starts a second client's sign-in.
The server counts failed sign-ins across its pods: 110 wrong passwords alternating between two front pods are banned at 101.
Through a proxy, the server counts each client: 110 failures for 110 clients ban nobody.
probe.js runs from a pod in data, paced under the server's 100-a-minute anonymous limit. blocked-ips.py lists and lifts bans as admin, and the test lifts any it causes. Each run uses its own account names, since the server's counts last a day.
Passes on kind with the webmail and server branches. On the rc.31 server, step 4 fails (the proxy is banned), which is the bug the server and chart PRs fix.
`lab/chart/ratelimit-test.sh`, after `frontends-test.sh`:
1. The webmail's own limit counts per replica: replica A refuses one client at attempt 201, and replica B still starts its sign-in.
2. The webmail counts the browser its proxy names, not the proxy: replica A still starts a second client's sign-in.
3. The server counts failed sign-ins across its pods: 110 wrong passwords alternating between two front pods are banned at 101.
4. Through a proxy, the server counts each client: 110 failures for 110 clients ban nobody.
`probe.js` runs from a pod in `data`, paced under the server's 100-a-minute anonymous limit. `blocked-ips.py` lists and lifts bans as admin, and the test lifts any it causes. Each run uses its own account names, since the server's counts last a day.
Passes on kind with the webmail and server branches. On the rc.31 server, step 4 fails (the proxy is banned), which is the bug the server and chart PRs fix.
The webmail's own limit counts per replica and per browser its proxy names;
the server counts failed sign-ins across its pods and bans after about 100;
through a proxy, each client counts, not the proxy. Bans the test causes are
lifted at the end (blocked-ips.py).
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
lab/chart/ratelimit-test.sh, afterfrontends-test.sh:probe.jsruns from a pod indata, paced under the server's 100-a-minute anonymous limit.blocked-ips.pylists and lifts bans as admin, and the test lifts any it causes. Each run uses its own account names, since the server's counts last a day.Passes on kind with the webmail and server branches. On the rc.31 server, step 4 fails (the proxy is banned), which is the bug the server and chart PRs fix.