Three years of smooth sailing. Now... Weirdness

Installed FreePBX a few years ago and it has been really reliable outside of a couple of issues with our SIP provider getting DDOS’d. We’ve got Yealink phones. FreePBX has been running on a Vultr instance. FreePBX is 15.0.38. Probably time to upgrade.

Anyways, a couple of days ago, we realized that calls to a bunch of extensions would go straight to voicemail. A bit of poking around and I found that restarting the phone seemed to fix the problem.

Today, I get an email notification that I’ve got a voicemail, but my phone didn’t ring and this person wasn’t the sort to just leave a voicemail. Look at my phone and see that it has no service. Run around the office and a few phones have the same problem.

I try restarting my phone and it doesn’t help. Check that the FreePBX instance is up. It is. The web interface seems fine. I can SSH to it. The load is a bit high. I figure it is worth updating FreePBX and do so through SSH (yum update -y; fwconsole moduleadmin upgradeall; fwconsole restart). That doesn’t make much difference so I do a fwconsole stop and restart the Vultr instance.

Still no difference. So I pull the POE on my phone for a couple of minutes and plug it back in. In the Asterisk logs I can see that my endpoint is reachable, but the phone still shows no service. As I browse the logs some more, possibly after restarting the phone again, the phone beeps like it has a voicemail. From the display, the phone looks fine.

But there’s still weirdness.

I can call out, but there’s no dialing sound. The call will connect and all is well, but sometimes the dialing sound plays after the call is connected.

Calls to most extensions are going to voicemail, including between internal extensions. So that’s no good either.

If I play my voicemail through my phone, it pauses forever before playing any of the voice queues (“Press 1 to play 1st message”, etc.). Once it starts, it’s fine.

Any ideas? I’m thinking it’s Vultr, but I’m having trouble pointing the finger at them.

Can you observe (and share) the sip registration flow to see if there are any oddities? I like sngrep for debugging these types of problems.

If the call to voicemail is hanging, watching the asterisk log might be helpful in seeing what it’s getting hung up on. Debugging sip on a call like this would likely be helpful too.

I poked around for a while, the web interface seemed sluggish, sometimes even timing out. That pushed me towards thinking that the problem was with Vultr, but I couldn’t see any networking problems. Pings from the office to FreePBX never showed a hiccup.

I stayed late continuing to poke around. I eventually reached the conclusion that I should grab the latest FreePBX backup and restore to a new VPS instance.

As I began preparing for that, the FreePBX web interface suddenly felt snappy. I picked up my phone and it behaved as it does normally. I ran around the office testing other phones and all seemed well again.

My guess was that Vultr was having issues. It really seems like the VPS was starved of resources.

I’ve definitely had many issues with Vultr instances that are on shared nodes being impacted by ‘noisy neighbors’ taking too many resources or being DDoSed and having it leak over and impact us. Glad it’s working well for you now.

ICYMI

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.