[google-cloud-sql-discuss] Re: App Engine to Cloud SQL connectivity problem

Thanks for the information, Yannick.

As of today 2017-12-07 I still experience the issue and it become noticeable. About 2% of all requests to my site are dropped due to this error. I did optimize the website to use less reads from DB and use Memcache more. I did clean the database from outdated entries to make it lighter. I also thought that probably I was reached the limits of my Cloud SQL level and I've upgraded to higher level, but none of these measures helped. Still the same percentage of dropped requests. Few observations:

1. The rate of "Lost connection" errors obviously has correlation with the amount of request per second to my website. When the request per second is less than 1/s I see quite few errors. When it gets higher to 4/s I see a lot of errors.

2. Errors have correlation with the latency. Almost every error entry in the log shows that request processing time was from 5s to 15s while cpu usage was quite small, so I believe all of this time it waits for connection to Cloud SQL.

3. Higher latency leads to creation of additional instances which actually doesn't solve the problem, but increases the cost for me.

4. Errors are started to appear in early November and most of the days 1-2% of requests were dropped. However, there were few days when it was just few errors of this kind while the amount of request were the same as for other days. Few months before website was working perfectly even with higher load.

Am Donnerstag, 30. November 2017 17:46:42 UTC+3 schrieb Yannick (Cloud Platform Support):

There was an issue with Google Cloud SQL high latency that has been resolved for all affected users as of Thursday, 2017-11-30 02:46 US/Pacific.

We will conduct an internal investigation of this issue and make appropriate improvements to our systems to help prevent or minimize future recurrence.

For any kind of billing-related issues you should contact the free Billing and Payments Support.

On Thursday, November 30, 2017 at 9:11:53 AM UTC-5, Andrew Karpushin wrote:

It's still an issue for me, and today it's worse than before. About 4-5 days ago i saw no errors, but starting from today i see hundreds and hundreds of them. Stackdriver reports from time to time that my site is down for few minutes. There is also DeadlineExceededError while app tries to connect to Cloud SQL (seen in the logs). Moreover, as a side effect, because the website doesn't respond in the appropriate time, App Engine creates 19 additional instances and I've got billed to their CPU-hours. My appid is "armorinspector" if that helps.

Am Montag, 13. November 2017 22:26:09 UTC+3 schrieb Yannick (Cloud Platform Support):

Hello Andrew, engineering is aware of this issue and a fix should soon be rolled out to address it.

On Saturday, November 11, 2017 at 12:05:38 PM UTC-5, Andrew Karpushin wrote:

In the past couple of days I started to see hundreds errors in the log like this one:

"Lost connection to MySQL server at 'reading initial communication packet', system error: 38"

Previously it was maybe one or two such errors per day, but now they appear every few minutes. Nothing was changed on my side and I believe something is wrong with Google SQL. Errors in the log started to appear at 2017-11-08. I can't reproduce it myself and most of the requests are handled properly. I have about 180k requests per day and several hundreds of them are now failing due to this error. Anyone else experience this?

--
You received this message because you are subscribed to the Google Groups "Google Cloud SQL discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-cloud-sql-discuss+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-cloud-sql-discuss/b2fa8d6d-5f4a-40ce-8213-1fa4533c46ec%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.