Closed
Bug 1344364
Opened 9 years ago
Closed 9 years ago
Need to open port 5432 (postgresql) for on relengapi webheads
Categories
(Infrastructure & Operations Graveyard :: NetOps: DC ACL Request, task)
Infrastructure & Operations Graveyard
NetOps: DC ACL Request
Tracking
(Not tracked)
RESOLVED
FIXED
People
(Reporter: garbas, Assigned: jbarnell)
Details
Hi,
RelengAPI is moving its database to PostgreSQL(Heroku). In Bug 1342251 I managed to get needed system packages installed, but now I would need to open port 5432 so I can connect to PostgreSQL database.
This are the logs I'm seeing
Mar 01 16:01:45 web1.stage.releng.webapp.scl3.mozilla.com relengapi: OperationalError: (psycopg2.OperationalError) could not connect to server: Connection timed out
Mar 01 16:01:45 web1.stage.releng.webapp.scl3.mozilla.com relengapi: #011Is the server running on host "ec2-184-73-202-229.compute-1.amazonaws.com" and accepting
Mar 01 16:01:45 web1.stage.releng.webapp.scl3.mozilla.com relengapi: #011TCP/IP connections on port 5432?
[root@web1.releng.webapp.scl3 ~]# traceroute -n -T -p 5432 ec2-184-73-202-229.compute-1.amazonaws.com
traceroute to ec2-184-73-202-229.compute-1.amazonaws.com (184.73.202.229), 30 hops max, 60 byte packets
1 * * *
2 * * *
3 * * *
4 * * *
5 * * *
6 * * *
7 * * *
8 * * *
...
| Reporter | ||
Comment 1•9 years ago
|
||
:justdave I got your contant from :coop. Did I assigned the ticket to the right group? Looking to be able to deploy this week.
Flags: needinfo?(justdave)
| Assignee | ||
Comment 2•9 years ago
|
||
Can you let me know if web1.releng.webapp.scl3 is configured to use the proxy servers? I'm just picking this up but that should get you up and running.
Flags: needinfo?(rgarbas)
| Assignee | ||
Updated•9 years ago
|
Assignee: network-operations → jbarnell
| Reporter | ||
Comment 3•9 years ago
|
||
:jbarnell I'm not really sure where/how to check this. Is there any mana/wiki page I can read about this?
Flags: needinfo?(rgarbas)
Flags: needinfo?(justdave)
Flags: needinfo?(jbarnell)
Comment 4•9 years ago
|
||
That's a question for someone in webops.
Flags: needinfo?(jbarnell) → needinfo?(smani)
Comment 5•9 years ago
|
||
No this system is not using the proxy servers yet, so a normal network ACL to open the port should be good.
Flags: needinfo?(smani)
libpq/pyscopg2 do not offer any known configurable option for using an HTTP proxy that I can find accessible at a Python level, so this application is probably not capable of using the outbound proxies without rewriting upstream DB code.
Comment 7•9 years ago
|
||
Also I believe this is the list of releng web servers that run the relengapi app and would likely need this ACL, feel free to correct me Rok:
web1.stage.releng.webapp.scl3.mozilla.com
web1.releng.webapp.scl3.mozilla.com
web2.releng.webapp.scl3.mozilla.com
relengwebadm.private.scl3.mozilla.com
Do the celery servers need the same port open Rok? If so, add celery[1-2].srv.releng.scl3.mozilla.com and celery1.stage.srv.releng.scl3.mozilla.com.
Rok, is this a permanent long-term ACL, or merely a temporary ACL as part of exiting to the cloud?
Flags: needinfo?(rgarbas)
| Reporter | ||
Comment 9•9 years ago
|
||
:atoll: this is needed until we migrate all traffic to new services. probably this is going to take us whole year, but majority of the trafic will be redirected in next month.
:ericz: celery servers also need the same port open.
Flags: needinfo?(rgarbas)
Comment 10•9 years ago
|
||
Okay. If it were possible to minimize the amount of time we keep this ACL open, that would be lovely, so please let us know as soon as you've migrated the bare minimum necessary for us to close it.
Comment 11•9 years ago
|
||
Sources:
web1.stage.releng.webapp.scl3.mozilla.com
web1.releng.webapp.scl3.mozilla.com
web2.releng.webapp.scl3.mozilla.com
relengwebadm.private.scl3.mozilla.com
celery1.srv.releng.scl3.mozilla.com and
celery2.srv.releng.scl3.mozilla.com and
celery1.stage.srv.releng.scl3.mozilla.com.
For the record, relengwebadm may already have the necessary ACL, but the workers surely don't. These could be tagged "Relengapi" or "Relengweb" if you need a shortname for the routers, and there might already be an ACL group leftover from the past for them.
| Reporter | ||
Comment 12•9 years ago
|
||
:atoll:
(1) I don't think I need the shortname.
(2) To confirm, I can already connect to postgresql from relengwebadm.
(3) For restoring back the ACL I created an Bug 1351073 to remind me.
Thank you.
Comment 13•9 years ago
|
||
Hi there,
just checking in on status of this bug? It's blocking the migration of releng services.
atoll, do you think you could give us an ETA of completion? Should this bug be assigned to you? I understand you have your own priorities so if we could just get an ETA, we can adjust our own goals and targets accordingly. Thanks in advance. Feel free to poke rok on vidyo or irc if it means less investigation and overall time for you. :)
Flags: needinfo?(rsoderberg)
Comment 14•9 years ago
|
||
Netops can help here.
James,
Comment #11 has the sources, they all need access to ec2-184-73-202-229.compute-1.amazonaws.com port 5432.
Jordan,
Please correct me if I'm wrong. Thanks!
Flags: needinfo?(rsoderberg) → needinfo?(jbarnell)
| Reporter | ||
Comment 15•9 years ago
|
||
To correct :fox2mike:
Staging env needs to be able to access ec2-184-73-202-229.compute-1.amazonaws.com on port 5432
This means the following sources:
- web1.stage.releng.webapp.scl3.mozilla.com
- celery1.stage.srv.releng.scl3.mozilla.com.
- relengwebadm.private.scl3.mozilla.com
Production environment needs to be able to access ec2-54-87-55-241.compute-1.amazonaws.com on port 5432
This means the following sources:
- web1.releng.webapp.scl3.mozilla.com
- web2.releng.webapp.scl3.mozilla.com
- celery1.srv.releng.scl3.mozilla.com and
- celery2.srv.releng.scl3.mozilla.com and
- relengwebadm.private.scl3.mozilla.com
At this time only `relengwebadm.private.scl3.mozilla.com` has the opened port 5432 and can acess both postgresql databases.
Comment 16•9 years ago
|
||
anything change here today, Monday, in and around 13:15-13:55 pacific?
a number of sql dependent services, notably archiver, tooltool, were taken down and we received 500s. e.g. Bug 1355205 - builds fail to download resources
garbas> jlund: connection to mysql is a problem
13:52:16 jlund: https://papertrailapp.com/groups/1421834/events?q=(program%3Arelengapi+OR+api.pub.build.mozilla.org)+-web1.stage+-celery1.stage
13:53:24
<jlund> Jordan Lund we should raise this in #moc and loop in webops
13:54:04
<garbas> it looks like mysql connection is back
13:54:11 maybe it was only temporary
13:56:44
<garbas> jlund: it shows tooltool is now serving
Comment 17•9 years ago
|
||
I looked in newrelic at api.pub.build.mozilla.org and it looks like it was a db auth issue rather than a flow, lots of tracebacks ending in
OperationalError: (_mysql_exceptions.OperationalError) (1045, "Access denied for user 'REDACTED'@'AAA.BBB.CCC.DDD' (using password: YES)")
Comment 18•9 years ago
|
||
Is this the cause of bug 1350273?
Comment 19•9 years ago
|
||
(In reply to Chris AtLee [:catlee] from comment #18)
> Is this the cause of bug 1350273?
Not 17 days ago, certainly. Did bug 1350273 reoccurred during today's 13:15-13:55 window?
Comment 20•9 years ago
|
||
hi jbarnell, can you confirm if you can resolve this bug and what the ETA is for doing so?
| Assignee | ||
Comment 21•9 years ago
|
||
pecking away at this (part 1):
[edit security policies from-zone srv to-zone untrust]
policy puppet--rsync { ... }
+ policy postgressql {
+ match {
+ source-address celery1.stage;
+ destination-address ec2-184-73-202-229.compute-1;
+ application PostgresSQL;
+ }
+ then {
+ permit;
+ }
+ }
[edit security zones security-zone untrust address-book]
address orange-antelope-static3.rmq.cloudamqp.com { ... }
+ address ec2-184-73-202-229.compute-1 184.73.202.229/32;
[edit applications]
application releng_to_puppet { ... }
+ application PostgresSQL {
+ protocol tcp;
+ destination-port 5432;
+ }
| Assignee | ||
Comment 22•9 years ago
|
||
Part 2
{primary:node1}[edit]
jbarnell@fw1.ops.releng.scl3.mozilla.net# show | compare
[edit security policies from-zone srv to-zone untrust]
policy postgressql { ... }
+ policy PostGres_prod {
+ match {
+ source-address [ celery1 celery2 ];
+ destination-address ec2-54-87-55-241.compute-1;
+ application PostgresSQL;
+ }
+ then {
+ permit;
+ }
+ }
[edit security zones security-zone untrust address-book]
address ec2-184-73-202-229.compute-1 { ... }
+ address ec2-54-87-55-241.compute-1 54.87.55.241/32;
Flags: needinfo?(jbarnell)
| Assignee | ||
Comment 23•9 years ago
|
||
Final bits
jbarnell@fw1.ops.scl3.mozilla.net# show | compare
[edit security policies from-zone private to-zone untrust]
policy inet-permit { ... }
+ policy releng_postgres {
+ match {
+ source-address relengwebadm.private.scl3;
+ destination-address [ ec2-184-73-202-229.compute-1 ec2-54-87-55-241.compute-1 ];
+ application postgres;
+ }
+ then {
+ permit;
+ }
+ }
[edit security policies from-zone webapp to-zone untrust]
policy mozreview_pulse { ... }
+ policy releng_postgres_stage {
+ match {
+ source-address web1.stage.releng.webapp.scl3;
+ destination-address ec2-184-73-202-229.compute-1;
+ application postgres;
+ }
+ then {
+ permit;
+ }
+ }
+ policy releng_postgres_prod {
+ match {
+ source-address [ web1.releng.webapp.scl3 web2.releng.webapp.scl3 ];
+ application postgres;
+ ## Warning: missing mandatory statement(s): 'destination-address'
+ }
+ then {
+ permit;
+ }
+ }
[edit security zones security-zone untrust address-book]
address contegix-118 { ... }
+ address ec2-54-87-55-241.compute-1 54.87.55.241/32;
+ address ec2-184-73-202-229.compute-1 184.73.202.229/32;
{primary:node1}[edit]
jbarnell@fw1.ops.scl3.mozilla.net#
| Reporter | ||
Comment 25•9 years ago
|
||
I've checked that all port 5432 is open on all staging and production relengapi servers.
:jbarnell thank you!
Status: NEW → RESOLVED
Closed: 9 years ago
Flags: needinfo?(jlund)
Resolution: --- → FIXED
Summary: Need to open port 5432 (postgresql) for on relenapi webheads → Need to open port 5432 (postgresql) for on relengapi webheads
| Reporter | ||
Comment 26•9 years ago
|
||
Hi,
I forgot to add "web3.releng.webapp.scl3.mozilla.com" to the list of production environments that have port 5432 open. Could we add this webhead to the list?
Here is the list of servers in production environment that needs access to ec2-54-87-55-241.compute-1.amazonaws.com on port 5432
- web1.releng.webapp.scl3.mozilla.com
- web2.releng.webapp.scl3.mozilla.com
- web3.releng.webapp.scl3.mozilla.com
- celery1.srv.releng.scl3.mozilla.com and
- celery2.srv.releng.scl3.mozilla.com and
- relengwebadm.private.scl3.mozilla.com
pinging :justdave: by :jbarnell:'s recommendation
Status: RESOLVED → REOPENED
Flags: needinfo?(justdave)
Resolution: FIXED → ---
Comment 27•9 years ago
|
||
justdave@fw1.ops.scl3.mozilla.net# show | compare
[edit security policies from-zone webapp to-zone untrust policy releng_postgres_prod match]
- source-address [ web1.releng.webapp.scl3 web2.releng.webapp.scl3 ];
+ source-address [ web1.releng.webapp.scl3 web2.releng.webapp.scl3 web3.releng.webapp.scl3 ];
CHG0011750
Status: REOPENED → RESOLVED
Closed: 9 years ago → 9 years ago
Flags: needinfo?(justdave)
Resolution: --- → FIXED
Updated•3 years ago
|
Product: Infrastructure & Operations → Infrastructure & Operations Graveyard
You need to log in
before you can comment on or make changes to this bug.
Description
•