About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
If rsyslog writes logs locally but nothing arrives at the central syslog server, compare the sender and receiver on four points: protocol, listening port, rule order and queue settings. The omfwd module's documentation gives the default protocol as udp, so an action that names no protocol sends UDP, and a receiver that listens only on TCP never sees it. A second trap is that an action's default retry count is 0, so messages are discarded while the remote server is down. The checks below find which one applies before any file is edited.
What the symptom means
rsyslog forwards messages with the built-in omfwd output module, which, in the module's words, "forwards logs to remote systems using UDP, TCP, or TLS". Its documented defaults are port 514 and protocol udp. The documentation says tcp "includes both legacy plain TCP syslog and RFC5425-based TLS-encrypted syslog".
The older short forms are easy to misread. The documentation shows *.* @@remote.example.com as the TCP form and *.* @remote.example.com:515 as the UDP form, and says these shorter forms "bypass queues".
On the receiving side, nothing arrives unless an input is loaded and listening. The basic structure page shows the pattern module(load="imtcp") followed by input(type="imtcp" port="514"), and the imudp page shows module(load="imudp") followed by input(type="imudp" port="514") for UDP.
Two more mechanics matter. Rules are "evaluated in order, top to bottom", and a stop "immediately halts processing for that message", so a forwarding action placed after a stop never sees those messages. And action queues "operate in direct (non-queueing) mode" by default, with action.resumeRetryCount defaulting to 0, which the actions page explains means "Failed actions discard messages."
Likely causes
| Cause | What you see | How to confirm |
|---|---|---|
| Protocol mismatch (omfwd default is UDP) | Sender shows no errors; receiver is silent | The action has no protocol and no @@; ss -lntu on the receiver shows only a TCP listener |
| Receiver has no input loaded or the port is wrong | Nothing listens on the port | ss -lntu on the receiver does not show the port |
| Firewall or network blocks the port | A TCP connection never establishes | ss -tn state established '( dport = :514 )' on the sender shows no connection |
Forward rule after a stop, or the filter does not match |
Local files get the message, the remote does not | Read the rule order in the configuration files |
| Configuration error, or change not loaded | The action never starts | rsyslogd -N1 reports an error; the file changed but rsyslog was not restarted |
| Remote server was down and the action discarded messages | A gap in remote logs, then forwarding resumes | The action sets no action.resumeRetryCount and no queue |
Prove the cause before changing anything
All of these only read or send one test message. Run them as root.
1. Check the configuration.
rsyslogd -N1
The manual says -N level does "a config check" and does "NOT run in regular mode". A good result reports no error. A bad result reports an error.
2. Find every forwarding action.
grep -rn -E 'omfwd|@@?[a-z0-9]' /etc/rsyslog.conf /etc/rsyslog.d/
Read each action. Note its target, port, protocol, any queue. settings, and which rules and stop statements come before it.
3. Send a test message.
logger "rsyslog forward test from web01"
The logger manual says it makes "entries in the system log". Look for that text in the local log and on the receiver.
4. On the sender, is a TCP connection open?
ss -tn state established '( dport = :514 )'
The ss manual documents -t, -n, state filters and dport expressions. A good TCP result shows one connection to the receiver. UDP has no connection, and the reliable forwarding paper notes that a down server "can be detected with plain TCP syslog and RELP, but not with UDP."
5. On the receiver, what is listening?
ss -lntu
The manual documents -l (listening), -n, -t and -u. Look for the port, and whether it is tcp, udp or both.
6. Read rsyslog's own messages.
journalctl -u rsyslog
The journalctl manual says -u shows messages for a unit. Look for configuration errors, and for messages about an action being suspended. The action.reportSuspension parameter exists so rsyslog will "report suspension and reactivation of the action".
Fix
Match the protocol and add a queue (sender)
First save the current file. Use the name of the file that holds the omfwd action:
cp -p /etc/rsyslog.d/90-forward.conf /etc/rsyslog.d/90-forward.conf.2026-11-22
Disruptive step: the restart after the edit stops rsyslog and starts it again, so local logging pauses for that moment and messages not yet written or sent can be lost. The reliable forwarding paper says queue.saveOnShutdown="on" persists data on an orderly shutdown, but that setting does not cover a kill, a crash or power loss.
Edit the action to the form the documentation shows for reliable forwarding, with your own target and port. The paper says rsyslog needs a working directory where it stores queue files, and that any local location works. Use a working directory that already exists; if your configuration already sets workDirectory, keep that one and leave the global(...) line out:
global(workDirectory="/var/spool/rsyslog")
action(
type="omfwd"
target="logs01.example.com"
port="514"
protocol="tcp"
action.resumeRetryCount="-1"
queue.type="LinkedList"
queue.filename="fwd_logs01"
queue.saveOnShutdown="on"
)
action.resumeRetryCount="-1" retries without end ("-1 means eternal"), and queue.filename enables disk assistance, so it must be unique in the configuration. Run the restart only if rsyslogd -N1 reports no error; if it does report one, stop and use the undo below. Then check the file and restart:
rsyslogd -N1 && systemctl restart rsyslog
To undo it: restore the file you saved above (the same file, under its original name) only if the saved copy exists, check it, then restart.
ls -l /etc/rsyslog.d/90-forward.conf.2026-11-22 && cp -p /etc/rsyslog.d/90-forward.conf.2026-11-22 /etc/rsyslog.d/90-forward.conf
rsyslogd -N1 && systemctl restart rsyslog
Load a listener (receiver)
First save the current file on the receiver:
cp -p /etc/rsyslog.conf /etc/rsyslog.conf.2026-11-22
Risk: the next edit opens a listening port that accepts log messages from any host able to reach it, and the restart pauses logging on the receiver. Restricting who can reach the port is a separate firewall change that is not covered here. The omfwd page advises TLS over untrusted networks.
Add the lines the documentation shows to /etc/rsyslog.conf, the file saved above:
module(load="imtcp")
input(type="imtcp" port="514")
Run the restart only if rsyslogd -N1 reports no error; if it does report one, stop and use the back-out below. Then check and restart:
rsyslogd -N1 && systemctl restart rsyslog
To back out: restore the file you saved above (the same file, under its original name) only if the saved copy exists, check it, then restart.
ls -l /etc/rsyslog.conf.2026-11-22 && cp -p /etc/rsyslog.conf.2026-11-22 /etc/rsyslog.conf
rsyslogd -N1 && systemctl restart rsyslog
Rule order
A forwarding action that sits below a stop is never reached for the messages that stop catches. The fix is to change the rule order or the filter.
First save the file that holds the stop or the forwarding action. If this is a file you already saved above, do not save it again, because the second copy would overwrite the saved original with the edited file. The name here is an example. If the action and the stop are in different files, save both, each with its own cp -p line:
cp -p /etc/rsyslog.d/50-default.conf /etc/rsyslog.d/50-default.conf.2026-11-22
Disruptive step: the next edit changes which messages reach which actions, and the restart after it stops rsyslog and starts it again, so local logging pauses for that moment. Messages that were written to a local file before may stop being written there if the order is changed carelessly.
Edit the file so that the forwarding action comes before the rule that ends in stop, or narrow the filter of that rule so it no longer catches the messages you want forwarded. Run the restart only if rsyslogd -N1 reports no error; if it does report one, stop and use the undo below. Then check the file and restart:
rsyslogd -N1 && systemctl restart rsyslog
To undo it: restore every file you saved above (the same files, under their original names) only if each saved copy exists, check the configuration, then restart.
ls -l /etc/rsyslog.d/50-default.conf.2026-11-22 && cp -p /etc/rsyslog.d/50-default.conf.2026-11-22 /etc/rsyslog.d/50-default.conf
rsyslogd -N1 && systemctl restart rsyslog
Example (illustrative): web01 (198.51.100.10) has the line *.* @logs01.example.com and the central server logs01 (203.0.113.20) loads only imtcp on port 514. rsyslogd -N1 is clean and logger appears in web01's local log. On logs01, ss -lntu shows a TCP listener on 514 and no UDP one. The single @ means UDP, so every message goes to a port nobody listens on. The engineer saves the file, replaces the line with the TCP action above (target logs01.example.com, protocol tcp), checks it, and restarts. ss -tn state established '( dport = :514 )' now shows one connection to 203.0.113.20, and a new logger message appears on logs01.
Confirm the fix
rsyslogd -N1reports no error.ss -tn state established '( dport = :514 )'on the sender shows a connection to the receiver (TCP).- A
loggertest message appears on the receiver. - Optional outage test, on the receiver only, and only in a window when pausing the central log server is acceptable. It is described after this list. The paper notes "actual spool files are only created if the remote server is down and there is no more space in the in-memory queue", so do not expect files for a short outage.
Optional outage test
First save the current state of the receiver's service:
systemctl is-active rsyslog > /root/rsyslog-receiver-state.2026-11-22
Disruptive step: stopping rsyslog on the receiver pauses reception of logs from every sender that uses it, not only the one you are testing, and what senders send during the pause depends on their own queue settings. Do it only if the saved state says active.
Stop rsyslog on the receiver, send a test message from the sender with logger, then start rsyslog again:
systemctl stop rsyslog
On the sender:
logger "rsyslog outage test from web01"
On the receiver:
systemctl start rsyslog
To undo it, start rsyslog again on the receiver:
systemctl start rsyslog
Then run systemctl is-active rsyslog and confirm it prints what the saved file holds. After the start, check that the messages sent during the pause arrive on the receiver.
What to watch afterwards
- Compare the received count with the sender's local log after any receiver maintenance.
- For high-volume or long outages, the omfwd page says to tune
queue.sizeand related parameters, and to consideromrelpwhere delivery must be guaranteed. - Re-run
rsyslogd -N1after every change to a file under/etc/rsyslog.d/.
Read next
Related articles you may want next.
- TroubleshootingESXi logs locally but nothing arrives at remote syslogESXi has local logs but the remote collector is empty. Compare logHost, forwarding errors, reload history and the ESXi 8.0 U2b port boundary.Read article
- TroubleshootingJournal logs filling the disk: how to limit /var/log/journal and vacuum it safely/var/log/journal is huge? Check journalctl --disk-usage, trim with --vacuum-size, and cap it with SystemMaxUse. What is lost, and what vacuum cannot touch.Read article
- HostingProvider administrator access in managed hosting: who can log in, how it is logged, and break-glass accountsWhat administrator access a managed hosting provider holds to a customer's servers, what Australian and Microsoft guidance says about controlling and logging privileged access, how emergency break-glass accounts should work, and the questions to ask before signing.Read article
Risks and limits
- These commands have not been run or lab-tested.
- Plain TCP syslog is "not fully reliable", in the paper's words; RELP is the module it names for stronger delivery.
- Distribution SELinux policy for non-standard ports is not covered.
semanage port -llists port definitions (semanage-port(8)); use your distribution's policy documentation for the right type. - The
journalctl -u rsyslogunit name can differ by distribution.
Sources
- omfwd: syslog Forwarding Output Module — rsyslog 8 documentation, no page date shown (site copyright 2008-2026), accessed 22 November 2026
- omfwd basic parameters — rsyslog 8 documentation, no date shown, accessed 22 November 2026
- Actions — rsyslog 8 documentation, no date shown, accessed 22 November 2026
- Understanding rsyslog Queues — rsyslog 8 documentation, no date shown, accessed 22 November 2026
- Reliable Forwarding of syslog Messages with Rsyslog — rsyslog 8 documentation, paper dated 2008-06-27, accessed 22 November 2026
- Basic structure of a configuration — rsyslog 8 documentation, no date shown, accessed 22 November 2026
- rsyslogd(8) — Debian Manpages, bookworm (rsyslog 8.2302.0), source updated 2025-01-08, accessed 22 November 2026
- logger(1) — man7.org, util-linux 2.43 development manual page, 2026-05-24, accessed 22 November 2026
- imudp: UDP Syslog Input Module — rsyslog 8 documentation, no date shown, accessed 22 November 2026
- ss(8) — Debian Manpages, bookworm, source updated 2023-05-22, accessed 22 November 2026
- journalctl(1) — Debian Manpages, bookworm, source updated 2025-03-06, accessed 22 November 2026
- systemctl(1) — Debian Manpages, bookworm, source updated 2025-03-06, accessed 22 November 2026
- semanage-port(8) — Debian Manpages, bookworm, source updated 2022-06-05, accessed 22 November 2026
This article was originally published on .