<div dir="auto">Hi.<div dir="auto">We faced similar issues during memory pressure. Some app leaked and caused intense swapping. You could check sar -B/sar -W if you have sysstat installed.</div><div dir="auto">WBR,</div><div dir="auto">Alexander Shevchenko</div><div dir="auto"><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">вс, 24 окт. 2021 г., 17:26 Christoph <<a href="mailto:cm@appliedprivacy.net">cm@appliedprivacy.net</a>>:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hello,<br>
<br>
we upgraded our debian BGP routers from debian buster to bullseye.<br>
<br>
On debian 10 they used the repo <a href="https://bird.network.cz/debian/" rel="noreferrer noreferrer" target="_blank">https://bird.network.cz/debian/</a><br>
now they use the bird2 package directly from the official debian repos,<br>
both repos contain BIRD version 2.0.7.<br>
<br>
<br>
After upgrading and rebooting we noticed that iBGP sessions are <br>
constantly dying every few minutes.<br>
<br>
<br>
The logs also show multiple of these log entries:<br>
<br>
<WARN> Kernel dropped some netlink messages, will resync on next scan.<br>
<WARN> Event ... took 31415 ms<br>
<WARN> I/O loop cycle took 31529 ms for 9 events<br>
<WARN> Netlink: File exists<br>
<br>
<br>
Especially the warning "Kernel dropped some netlink messages" was not <br>
new but we previously solved them with these sysctl.conf settings:<br>
<br>
<br>
# solves the "Kernel dropped some netlink messages, will resync on next <br>
scan." BIRD Warnings<br>
# <a href="https://bird.network.cz/pipermail/bird-users/2017-September/011542.html" rel="noreferrer noreferrer" target="_blank">https://bird.network.cz/pipermail/bird-users/2017-September/011542.html</a><br>
# (we did not change the wmem_default since this is apparently not needed)<br>
net.core.rmem_max=4194304<br>
net.core.rmem_default=4194304<br>
net.core.wmem_max=4194304<br>
<br>
<br>
The most obvious change is the linux kernel version:<br>
4.19.208 (buster)<br>
5.10.70 (bullseye)<br>
<br>
To avoid constantly dying iBGP sessions we increased the keepalive and <br>
hold timers.<br>
Are you observing similar issues on your debian 11 systems?<br>
Does the newer kernel need higher rmem values?<br>
<br>
best regards,<br>
Christoph<br>
</blockquote></div>