英文:
All App Engine Instances restarted on 25.12. and 31.12
问题
以下情况:
我们有多个使用Google App Engine的GCP项目,而在staging/production环境中,我们使用On-Prem-Varnish进行缓存,位于我们的实例前面。
由于Varnish,我们只运行每个App Engine服务的单个实例,这在负载方面完全正常。
发生的情况:
于2019年12月25日14:35 UTC和2019年12月31日21:45 UTC,所有我们的App Engine实例同时重新启动,导致了大约2-3分钟的停机时间。
对我来说,看起来实例是由Google重新启动的,但我找不到任何证据,除了日期和所有实例同时重新启动的事实。
问题:
如果这是Google的事情(例如,App Engine更新或其他什么),有人能指引我查找信息资源吗?
如果不是:如果我们运行多个实例,是否可以避免这种停机时间?我可以想象,如果Google更新GAE,他们不会一次关闭一个服务的所有实例。
编辑:
GAE实例正在运行在灵活环境中。
英文:
Following situation:
We have multiple GCP Projects using Google App Engine and in staging/production we do have a On-Prem-Varnish for caching in front of our instances.
Because of the Varnish we are only running a single instance of every App Engine Service which works totally fine regarding the load.
This happened:
On 25.12.2019 14:35 UTC and 31.12.2019 21:45 UTC all of our App Engine instances restarted at the same time which lead to a downtime of about 2-3 min.
For me it looks like the instances have been restarted by Google but I can't find any evidence for that, except for the dates and the fact, that all instances restarted at the same time.
Question:
If it was a Google thing (e.g. App Engine Update or something) can someone point me to the information resource?
If not: Could we have avoided this downtime if we had more than one instance running? I could imagine that if Google updates GAE, that they won't take down all instances of one service at a time.
EDIT:
The GAE instances are running on flexible environment.
答案1
得分: 0
我猜测您的GAE服务正在使用灵活环境,如果是这样,每周重新启动是预期的,并已记录。
这些重新启动是按服务进行的。由于您提到所有的服务只有一个实例,这些重新启动的确在那个时候引起了停机。谷歌建议至少运行两个服务实例,以避免这些停机。
英文:
I'd guess your GAE services are under the Flexible environment, in which case a weekly restart is expected and documented.
These restarts are performed on a per-service basis. Since you mentioned all your services have only one instance, these restarts indeed cause a downtime during that time. Google recommends to have a least 2 instances of a service running to avoid these downtimes.
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库,让每个人都能够通过互相帮助和分享经验来进步。
评论