MySQL AWS RDS 实例升级停机时间
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/4832459/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
AWS RDS instance upgrade down time
提问by Chuck Morris
I have a few questions in regards to upgrading the RDS instance.
我有几个关于升级 RDS 实例的问题。
- What is the downtime when upgrading the instance from let's say small to large. Is the downtime relatively similar when you go and change any instance type(small, large, xlarge) or are there determining factors such as database size that alter the timing.
- Can anyone share a technique of how to upgrade the instance type avoiding the downtime using RDS? Is that even possible in RDS. It doesn't have to be in great detail just some cliff-notes/big-picture stuff.
- Is there down time when you allocate more disk space?
- 将实例从小到大升级时的停机时间是多少。当您更改任何实例类型（小型、大型、超大型）时，停机时间是否相对相似，或者是否存在改变时间的数据库大小等决定性因素。
- 任何人都可以分享如何使用 RDS 升级实例类型避免停机的技术吗？这在 RDS 中甚至可能吗？它不必非常详细，只是一些悬崖笔记/大图。
- 分配更多磁盘空间时是否会出现停机时间？
采纳答案by Dan Grossman
I don't think this is an on-topic question for StackOverflow at all, but some information anyway:
我认为这根本不是 StackOverflow 的主题问题，但还是有一些信息：
It's significant and depends on the size of the database. I've had it take an hour or more some times. I've also had creating snapshots, restoring from snapshots, and multi-az creation take around two hours before.
It depends on how you have things configured now. If you have Multi-AZ already enabled, then an instance upgrade will actually occur on the slave, then a failover will occur, then the new slave is updated. This results in about 1 or 2 minutes of actual downtime. The instance upgrade on the slave usually takes around 10 to 20 minutes, but there is no downtime in this setup. Note that when it does the failover, Amazon does a DNS swap internally so that your RDS endpoint points to the right machine, so you may have to restart your web processes that point to the DB so that they reconnect to the DB and pull in the new IP from a new DNS lookup.
它很重要，取决于数据库的大小。我有几次需要一个小时或更长时间。我还创建了快照、从快照恢复以及创建多 az 大约需要两个小时。
这取决于您现在如何配置事物。如果您已经启用了多可用区，那么实例升级实际上会在从属上发生，然后会发生故障转移，然后更新新的从属。这会导致大约 1 或 2 分钟的实际停机时间。从站上的实例升级通常需要大约 10 到 20 分钟，但在此设置中没有停机时间。请注意，在进行故障转移时，Amazon 会在内部进行 DNS 交换，以便您的 RDS 端点指向正确的机器，因此您可能必须重新启动指向数据库的 Web 进程，以便它们重新连接到数据库并拉入来自新 DNS 查找的新 IP。
回答by Alastair
db.t1.micro> db.m1.small: 8m30s
db.t1.micro> db.m1.small: 8m30s
Engine: mysql
Storage: 6GiB
Backups: Yes
Multi A-Z: No
The size/type of the database do appear to affect the downtime significantly.
数据库的大小/类型确实会显着影响停机时间。
回答by DAiki
1, From personal experience it takes just short of one hour, to be precise 57min for 15 GB instance goign from small to large. Which I did not expect to be that long to be honest. update: just learned that switching point in time backup before upgrade speeds up the process significantly
1、根据个人经验，15GB实例从小到大只需要1小时，准确地说是57分钟。老实说，我没想到会这么久。更新：刚刚了解到在升级前切换时间点备份可以显着加快进程
2, I would say creating MULTI AZ before doing upgrade would do the trick, hopefully that does not have downtime as well. Question is do they allow upgrading one without other...
2，我会说在升级之前创建 MULTI AZ 可以解决问题，希望不会有停机时间。问题是他们允许升级一个没有其他...
3, yes, but I'm not 100% sure though
3，是的，但我不是 100% 确定
