We have a node.js web server that makes some outgoing http requests to an external API. It's running in docker using dokku.
After some time of load (30req/s) these outgoing requests aren't getting responses anymore.
Here's a graph I made while testing with constant req/s:
incoming and outgoing is the amount of concurrent requests (not the number of initialized requests). (It's hard to see in the graph, but it's fairly constant at ~10 requests for each.)
response time is for external requests only.
You can clearly see that they start failing all of a sudden (hitting our 1000ms timeout).
The more req/s we send, the faster we run into this problem, so we must have some sort of limit we're getting closer to with each request.
I used netstat -ant | tail -n +3 | wc -l on the host to get the number of open connections, but it was only ~450 (most of them TIME_WAIT). That shouldn't hit the socket limit. We aren't hitting any RAM or CPU limits, either.
I also tried running the same app on the same machine outside docker and it only happens in docker.
It could be due to the Docker userland proxy. If you are running a recent version of Docker, try running the daemon with the --userland-proxy=false option. This will make Docker handle port forwarding with just iptables and there is less overhead.
Related
We recently had an issue in our rabbitmq server , where it was unable to accept new connections and dropping those TCP connections .
we didn’t saw any spike in our channels or consumers .
Socket Descriptors(SD) and erlang process shoot up in short span of time causing Rabbit MQ to get stuck and no new connections get established post that.
We do not see any significant increase in channels, connections or consumers to establish a link between the sudden increase in SD and erlang Processes.
RMQ VERSION: 3.7.14
Erlang version: Erlang 21.3.8.1
RMQ running on Kubernetes as a stateful set .
RMQ erlang process spike .
Socket used.
Post restarting the server its working fine , but its resurfacing again .
I suggest you to check server's half-open connections. Seems you can have that kind of a situation if you have an aggressive reconnects from clients side. They create connections, and reconnect again and again.
Also, even if you have the same amount of consumers, there can be increased amount of publishers.
So, my suggestion here - to check logs and metrics on reconnects to rabbitmq.
We sometimes use Python scripts to spin up and monitor Kubernetes Pods running on Google Kubernetes Engine using the Official Python client library for kubernetes. We also enable auto-scaling on several of our node pools.
According to this, "Master VM is automatically scaled, upgraded, backed up and secured". The post also seems to indicate that some automatic scaling of the control plane / Master VM occurs when the node count increases from 0-5 to 6+ and potentially at other times when more nodes are added.
It seems like the control plane can go down at times like this, when many nodes have been brought up. In and around when this happens, our Python scripts that monitor pods via the control plane often crash, seemingly unable to find the KubeApi/Control Plane endpoint triggering some of the following exceptions:
ApiException, urllib3.exceptions.NewConnectionError, urllib3.exceptions.MaxRetryError.
What's the best way to handle this situation? Are there any properties of the autoscaling events that might be helpful?
To clarify what we're doing with the Python client is that we are in a loop reading the status of the pod of interest via read_namespaced_pod every few minutes, and catching exceptions similar to the provided example (in addition we've tried also catching exceptions for the underlying urllib calls). We have also added retrying with exponential back-off, but things are unable to recover and fail after a specified max number of retries, even if that number is high (e.g. keep retrying for >5 minutes).
One thing we haven't tried is recreating the kubernetes.client.CoreV1Api object on each retry. Would that make much of a difference?
When a nodepool size changes, depending on the size, this can initiate a change in the size of the master. Here are the nodepool sizes mapped with the master sizes. In the case where the nodepool size requires a larger master, automatic scaling of the master is initiated on GCP. During this process, the master will be unavailable for approximately 1-5 minutes. Please note that these events are not available in Stackdriver Logging.
At this point all API calls to the master will fail, including the ones from the Python API client and kubectl. However after 1-5 minutes the master should be available and calls from both the client and kubectl should work. I was able to test this by scaling my cluster from 3 node to 20 nodes and for 1-5 minutes the master wasn't available .
I obtained the following errors from the Python API client:
Max retries exceeded with url: /api/v1/pods?watch=False (Caused by NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at>: Failed to establish a new connection: [Errno 111] Connection refused',))
With kubectl I had :
“Unable to connect to the server: dial tcp”
After 1-5 minutes the master was available and the calls were successful. There was no need to recreate kubernetes.client.CoreV1Api object as this is just an API endpoint.
According to your description, your master wasn't accessible after 5 minutes which signals a potential issue with your master or setup of the Python script. To troubleshoot this further on side while your Python script runs, you can check for availability of master by running any kubectl command.
I am hosting a web application on Amazon's AWS Servers. I am currently in the process of load testing the application with JMeter. My main problem seems to be that when I go through an Elastic Load Balancer (ELB) to hit the Amazon server's rather than hitting the servers directly - I seem to hit a cap in my throughput.
If I hit my web application directly - for each server I am able to achieve a throughput of 50 RPS per server.
If I hit my web application via Amazon's ELB - I am only able to achieve a max throughput of 50 RPS (total)
I was wondering if anyone else has experienced similar behavior when load testing using Jmeter via Amazon's ELB.
For more context my web application is a REST application which allows users to download content (~150 kb) via HTTP requests.
I am running Jmeter with the following flag "-Dsun.net.inetaddr.ttl=0" and running it with 10 threads. I have tried running these tests with multiple clients on different machines.
Thanks for any help in advance.
Load balancers may be tricky to test as they may have different mechanisms of orchestrating traffic depending on origin. The most commonly used approach to distinguish origin of the request and redirect it to the same host, which served previous request is a cookie. You can look into HTTP Cookie Manager to correctly manipulate your cookies and make sure than you have different ones for each testing thread or thread group (depending on your use case). Another flaky area is origin host IP. You may require to bind each testing thread to different IP address in order to hit different servers behind the load balancer. There can be also some issues with DNS in regards to Amazon LBs. useful guide on how to test Amazon ELBs
Most probable cause would be DNS caching by jmeter. ELB returns IPs of additional servers depending on how autoscaling is set but JMeter does not use these additional servers. This problem can be solved by ensuring that Jmeter does not cache DNS results...
The ELB is a name, not IP, and can suffer from DNS caching. Make sure you use "-Dsun.net.inetaddr.ttl=0" when starting JMeter
http://wiki.apache.org/jmeter/JMeterAndAmazon
A really late response, and slightly different than the original question, but I hope this can help others as it took me a while to get it all straight. My original problem was not reduced throughput as a result of the ELB, but the introduction of HTTP 503 errors. Actually, the ELB increased my throughput as compared to querying the web application directly, though even with 1 hour tests, the results were sporadic to say the least.
First, the ELB has 2-staged load balancing going on. The first load balance is across the ELB's themselves. That's done by associating multiple IP addresses to the hostname provided by AWS for the ELB you provision. The second is then, of course, across your application instances behind the ELB.
Without trying to offend the SO gods, this is a really helpful article.
https://blazemeter.com/blog/dns-cache-manager-right-way-test-load-balanced-apps
The most helpful information in there was to use the DNS Cache Manager module in JMeter. This will query multiple DNS servers, and wipe out your DNS cache.
I implemented that module and then setup Wireshark, filtering on the two IP addresses belonging to the ELB hostname and sure enough, it was querying both IP addresses, though clearly favored one over the other.
That didn't make a big difference, at least not over short tests.
The real difference (2-3 times more throughput) came when I tweaked the ELB health settings. I initially had a high error rate, however after reducing the unhealthy threshold and the interval between health checks, my error rates dropped dramatically.
Additionally, whereas all my other tests had been 60 - 90 minutes in duration, this one was 8 hours. I started out with decent throughput and it then quickly dropped (by about 2/3). After about 20 minutes or more, the throughput then started ticking back up and by the end of the test, it had sustained throughput of about 5 times what I was getting without the ELB (which was similar to what the throughput was when it dropped shortly after beginning this test).
When we have a cluster of machines behind a load balancer (lb), generally hardware load balancer have persistent connections,
Now when we need to deploy some update on all machines (rolling update), the way to do is by bringing one machine Out of rotation, looks for no request sent to that server via lb. When the app reached no request state then update manually.
With 70-80 servers in picture this becomes very painful.
Can someone have a better way of doing it.
70-80 servers is a very horizontally scaled implementation... good job! Better is a very relative term, hopefully one of these suggestions count as "better".
Implement an intelligent health check for the application with the ability to adjust the health check while the application is running. What we do is have the health check start failing while the application is running just fine. This allows the load balancer to automatically take the system out of rotation. Our stop scripts query the load balancer to make sure that it is out of rotation and then shuts down normally which allows the existing connections to drain.
Batch multiple groups of systems together. I am assuming that you have 70 servers to handle peak load. This means that you should be able to restart several at a time. A standard way to do this is to implement a simple token granting service with a maximum of 10 tokens. Have your shutdown scripts checkout a token before continuing.
Another way to do this is with blue/green deploys. That means that you have an entire second server farm and then once the second server farm is updated switch load balancing to point to the new server farm.
This is an alternate to option 3. Install both versions of the app on the same servers and then have an internal proxy service (like haproxy) switch the connections between the version of the app that is deployed. For example:
haproxy listening on 8080
app version 0.1 listening on 9001
app version 0.2 listening on 9002
Once you are happy with the deploy of app version 0.2 switch haproxy to send traffic to 9002. When you release version 0.3 then switch load balancing back to 9001 etc.
I have a bug where a long running process works fine for the first few days, but then a query to redis reaches the 45 second timeout I have set. That is, if redis was totally down my program would just crash, but it doesn't. It waits and waits (45 seconds) timesout and tries again for another 45 seconds over and over.
If I stop the process and re-start it, everything is fine again for another few days.
This is running on ec2 with Elastic Load Balancing with my process on a different box than redis.
I need to re-create this situation on my local development environment. How can I not kill my local redis, but rather put it into a state where reads will timeout?
Maybe turn off the port? This might be interpreted as connections refused/down.
Perhaps put another non-redis app on said port and just have it not respond. In other words, accept incoming connections but don't respond. You could probably write a simple app that accepts TCP connections and then does nothing in the language of your choice, and have it start on the Redis port in order to test this situation.