Closed kyrieb-ekat closed 1 month ago
Have this when starting Rodan (develop branch) locally as well. Still trying to find where it comes from.
Related error msg:
Failed to load resource: the server responded with a status of 401 (Unauthorized)
http://localhost/api/auth/me/
Update: Still not sure why it is called... Also this login page does not work at all, i. e., you cannot input your username and pwd to complete authentication and expect it to disappear. I tried the rodan admin account (can be found on internal wiki) and it didn't work. I don't know if any account can work for this authentication box!
Found the related old PR #1076 And there is already an issue (still open) #600
The new Rodan prod (which is supposed to function properly) is having the same issue! It didn't happen before, even though I basically replicated exactly the same steps when setting up everything 🥀🥀🥀
It looks like a serious problem with both staging (develop) and production (master)!!! If anyone has any idea please let me know 🙏🙏🙏
One message from nginx container that seems related:
rodan_nginx.1.332fayd5ur38@prod-rodan-swarm-debian
2024/05/31 17:54:16 [warn] 423#423: *1273 an upstream response is buffered to a temporary file /var/cache/nginx/proxy_temp/3/08/0000000083 while reading upstream, client: 10.0.0.2, server: ,
request: "GET /rodan_client.min.js HTTP/1.0", upstream: "http://10.0.1.20:80/rodan_client.min.js",
host: "rodan2.simssa.ca", referrer: "https://rodan2.simssa.ca/"
Update: I have no issue using Safari but have this popup window constantly on chrome.
~I tried on the old mac mini I used to debug Rodan earlier and with latest codes but images from 4 months ago, there is no popup window issue at all. But we did not have much change to backend since my last commit. I guess it might be a network issue between containers.~
ddmal/iipsrv nightly dd41d2f70774 4 months ago 21.5MB
ddmal/nginx nightly 4d9ae548e0ac 4 months ago 213MB
ddmal/postgres-plpython nightly eb22c4407980 4 months ago 575MB
ddmal/rodan-gpu-celery nightly 5731084bd3e9 4 months ago 7.27GB
ddmal/rodan-client nightly cb9ea3495cce 4 months ago 50.5MB
ddmal/rodan-main nightly 8d2df5ae8d7d 4 months ago 10.1GB
ddmal/rodan-python3-celery nightly aa2daa7fb6fb 4 months ago 7.61GB
redis alpine 287766fc4fcf 4 months ago 41MB
rabbitmq alpine e518f1524750 4 months ago 136MB
Update Current Rodan prod server (rodan2.simssa.ca) fails on training (might be a problem on my end or a GPU-related thing) but if we skip the upstream training job and go into IC, we just need to enter login credential once and it will be gone. (But it is a tiny server, so everything will be slow.)
I just tested it on staging, and it works the same way. According to PR #600, this is expected.
It's probably better to use an account with an email address that's exclusively associated with this account. It also has to be the account you log in with. (Tested multiple times with different accounts locally and on servers.)
However, this popup window did appear on the prod Rodan homepage and didn't go away easily. There might be an issue related to this, but the use of IC does not seem to be affected. I suggest we keep this issue open, even though IC should have no serious problems that prevent users from getting the work done.
https://github.com/DDMAL/Rodan/assets/70364944/bb714721-2bfe-4a8a-8f2d-9243213c6552
![]()
The new Rodan prod (which is supposed to function properly) is having the same issue! It didn't happen before, even though I basically replicated exactly the same steps when setting up everything 🥀🥀🥀
It looks like a serious problem with both staging (develop) and production (master)!!! If anyone has any idea please let me know 🙏🙏🙏
One message from nginx container that seems related:
rodan_nginx.1.332fayd5ur38@prod-rodan-swarm-debian 2024/05/31 17:54:16 [warn] 423#423: *1273 an upstream response is buffered to a temporary file /var/cache/nginx/proxy_temp/3/08/0000000083 while reading upstream, client: 10.0.0.2, server: , request: "GET /rodan_client.min.js HTTP/1.0", upstream: "http://10.0.1.20:80/rodan_client.min.js", host: "rodan2.simssa.ca", referrer: "https://rodan2.simssa.ca/"
Update: I have no issue using Safari but have this popup window constantly on chrome.
This is related to postgre
error LOG: incomplete startup packet
Modified inside postgre (might not be related):
nano /var/lib/postgresql/data/pg_hba.conf
and had the last few lines as
# Allow replication connections from localhost, by a user with the
# replication privilege.
#local replication someadmin trust
#host replication someadmin 127.0.0.1/32 trust
#host replication someadmin ::1/128 trust
#host replication someadmin 10.0.2.55/32 trust
host all all all md5
I cannot register for a new user etc. at all on the homepage using my laptop (on Chrome and Safari) but I was able to do it with my iPhone (Safari) without seeing the popup window. Last time it was the same but I thought it was because of Safari which is probably wrong. However, once I register (on iPhone), I can log in and use other functions as expected.
This is all on production. Staging is not affected by this. Even on prod, IC has no login popup window issue.
Video shows the issue. Using the staging server, in chrome.
Essentially, when you open the interactive classifier you will be prompted with a login box. If you input your credentials it disappears and immediately reappears. If you hit cancel you gain about five seconds before it reappears. In the past this has happened when copying an error code, see #1147 .
https://github.com/DDMAL/Rodan/assets/74431986/875ef95a-815f-440c-91a1-299ccc82f3a7