Multiple issues after upgrading (please help)

[Former user]

I just upgraded to the latest version (2.0.14 PRO), and I'm seeing multiple problems:

• Internal server error (500) on delivery log pages (f.e. /admin/log/2018-07-26)
  • Which log file can I check for errors?
• On the Server connection diagnostics page (/admin/server/connection), I see 3 red errors:
  • Port 80 (HTTP was not redirected to HTTPS (mail.example.com -> http://mail.example.com/webmail/)
    • This may just be because we turn HTTPS off?
  • Port 443 (Connection refused)
    • Also because HTTPS off?
  • Port 4190 (Connection refused)
    • Maybe just because this is optional?
• There may be more problems, I'm still investigating

[Glen A.]

I think I found the error:

== [2018-07-27 13:28:12] request.CRITICAL: Uncaught PHP Exception Exception: "Can't parse row "2018-07-27 00:00:06 [e-mail] delivery 85854: success: 40e9:a61b:400c:c09::1a_TLS_transmitted_message_accepted./Remote_host_said:250_2.0.0_OK_1532642406_h2-v6si1938066wmg.128-gsmtp/ "" at /opt/admin/src/ProBundle/Logs/TransactionLogs.php line 109 {"exception":"[object] (Exception(code: 0): Can't parse row \"2018-07-27 00:00:06 [e-mail] delivery 85854: success: 40e9:a61b:400c:c09::1a_TLS_transmitted_message_accepted./Remote_host_said:_250_2.0.0_OK_1532642406_h2-v6si1938066wmg.128-_gsmtp/\n\" at /opt/admin/src/ProBundle/Logs/TransactionLogs.php:109)"} [] ==

[Glen A.]

Regarding issue 4, mail is being sent from 172.17.0.1 (AFAIK), and we have 172.17.0.0/16 listed under Relay networks.

[Scott MacDonald]

If you are running a firewall (UFW, Firewalld, or simply iptables) you need to ensure that ports 80 and 443 are opened so that the docker container can loopback onto itself. The docker container does open these ports but only for external inbound connections -- LE certificates will fail without these ports opened as the container attempts a test before proceeding with acquiring the certificates.

What version were you previously using? As there is an incompatibility with log files in any version pre-2.x.

[Glen A.]

Hi Scott,

If you are running a firewall ...

Is this something new? Everything was working before the upgrade.

I desperately need to get the relaying issue resolved ASAP. Do you know which log would show information about relaying failures?

What version were you previously using? As there is an incompatibility with log files in any version pre-2.x.

I think a pre-2.x release. It would have been nice if this was mentioned in the changelog.

So the code will fail to parse all existing logs (including today's logs), and I'll have to wait until tomorrow to see new logs, I guess.

[Glen A.]

Where is the Haraka log file?

I see the allowed relay IPs/ranges in: /opt/haraka-smtp/config/relay_acl_allow

Should there also be a relay_acl_allow file in the haraka-submission/config folder?

[Glen A.]

I just copied the relay_acl_allow file across to the submission folder, and it appears to have fixed the relaying issue. Is this a bug? Should Poste.io be writing to that folder as well?

[Scott MacDonald]

Poste.io V2 was re-architected which invalids viewability of the logs when moving from V1.0.7. There are a few closed tickets on this issue but there should perhaps be a more descriptive note that older logs will not be viewable via the webUI. I upgraded from 1.0.7 to 2.0.1 without any functional issues back at a the end of April. I've not encountered needing to add the docker container IP to the relay networks. Are you doing some custom configuration?

Have you attempted to stop/remove and recreate the container to see if it's just a glitch with having just upgraded versions? It's a big jump from 1.0.7 to 2.0.14 if that was the upgrade path.

What's your current status?

[SH]

1. Internal server error (500) - as stated it seems that old logs hit new UI, I am not sure why this happened because new UI uses different directory (/data/log/delivery), anyway please stop container + rename/delete old delivery folder and start again. Old logs are not viewable at 2.x anyway
2. these errors are informative only - Poste can't predict how it will be used (eg. behind reverse proxy) but you should use redirect and TLS - It is OK if you don't but you should know why.
3. I am not sure that I correctly understand. You have domain "localhost.mail.example.com" and it is not localy delivering?
4. Confirmed, it is bug, you can temporary use port 25 and it should work as intended

[SH]

ref #410 push relay subnets to submission also

→ <<cset a02d35e5e69e>>

[Glen A.]

@CyberMac,

Are you doing some custom configuration?

Not that I know of, but this has now been confirmed as a bug anyway.

Have you attempted to stop/remove and recreate the container to see if it's just a glitch with having just upgraded versions?

I didn't, but as mentioned above, this has been resolved.

What's your current status?

I'll post an update with my reply to @analogic.

Thanks.

[Glen A.]

@analogic,

1. I'll just ignore old logs, unless that'll cause other issues? (maybe related to stats?)

2. I'm not too worried about these errors. I didn't configure the server myself (I'm not even a sysadmin), so I'll assume that it's for a good reason (for now).

3. The emails are being delivered, but the to/from addresses changed after the upgrade. Again, it's not a big deal, I just wanted to mention it.

4. Cool, it only took me about 5 hours to figure that out, LOL.

There is one new issue:

• One of our email accounts (database@...) which redirects to 9-10 addresses, is not working. The delivery to database@ is showing up, but not the delivery to the individual redirect addresses. Other redirect accounts appear to be working though. Any ideas?

[SH]

1. it should not be problem stats are stored entirely at redis at 2.x
2. please go through logs - there should be explanation what is going on. 2.x version has entirely different smtp&delivery server (qmail+qpsmtpd -> haraka)
3. same

If you don't understand whole log, please send me them to [e-mail]

[SH]

fix #410 raise sieve_max_redirects

→ <<cset 367f5ae2affb>>