My Tech-Notes

Common reasons for high Koha server load

1. Bots hitting OPAC

• Solved by blocking crawlers with robots.txt + X-Robots-Tag.

2. Heavy API usage

• Some integrations or scripts might hit /api/v1/... endpoints repeatedly.

• You can check logs:

tail -f /var/log/koha/yourinstance/opac-access.log tail -f /var/log/koha/yourinstance/intranet-access.log

3. Plack misconfigured

• Not enough or too many Plack workers.

• Misaligned with server RAM/CPU.

4. Indexing jobs running too often

• Zebra or Elasticsearch indexing can spike usage.

---

Solutions : Here’s what you can do safely:

---

Step 1: Tune Plack workers

Koha uses plack for faster Perl web serving. Too many or too few workers can cause high load or slow performance.

Check current workers:

cat /etc/koha/sites/yourinstance/koha-conf.xml | grep -i prefork

Example section:

<plack_max_requests>50</plack_max_requests>
<plack_workers>2</plack_workers>

Basic rule:

• For 2 CPU cores, use 2 workers.

• For 4 CPU cores, try 3–4 workers.

---

Step 2: Limit bots

Use:

• robots.txt → stop OPAC crawling.

• fail2ban or iptables → block aggressive IPs.

• Apache mod_evasive → throttle abusive requests.

---

Step 3: Optimize Zebra / Elasticsearch

If you use Zebra:

• Don’t run indexing too frequently.

• Run at low-traffic times:

sudo koha-rebuild-zebra -v -f yourinstance

---

Step 4: Check cron jobs

Sometimes cron jobs hammer the server. Look:

crontab -l -u koha

See if koha-run-backups, koha-index-daemon, or cronjobs are too frequent.

---

Key tip

If you don’t need Koha’s REST API externally, block it in Apache or firewall.

Example Apache:

<Location /api>
   Require ip YOUR.INTERNAL.IP.RANGE
</Location>

Or block it completely with:

RedirectMatch 403 ^/api