Redian新闻
>
goodbug 以你的经验, EC2 instance 有多大概率,会自己down?
avatar
goodbug 以你的经验, EC2 instance 有多大概率,会自己down?# Programming - 葵花宝典
HQ
1
港府+内地资金,跟索洛斯从8月初斗到9月份,中间索洛斯把8月的空仓转到9月,最后抗不
住了
.
我们所谓的胜利是惨胜,索洛斯方的情况未知
从昨天的架势来看,大有当时的状态了.
不过这次无论规模还是情况,都要复杂得多.
因为外资有在美国的etf和在香港建立空仓获利的途径
avatar
l*0
2
谢谢,请问各位大侠指点。
avatar
t*1
3
今天终于收到797了,漫长的waiver也结束了。从本版获得了很多信息,作为回馈分享
我的timeline。希望大家的waiver都顺顺利利!
2013.1.30 寄材料到所属的领事馆
2013.3.19 寄材料到DOS
2013.3.27 中国大使馆发出NOL
2013.4.3 DOS显示收到材料
2013.4.9 DOS显示收到NOL
2013.4.22 DOS sent FR
2013.4.30 USCIS寄出I-797C
2013.5.3 USCIS寄出797
avatar
z*o
4
My parents have legally stayed in the U.S for about a year. Their I-94 was
extented to 1 year. My parents want to come to U.S again right after going
back to China. How long do they have to wait until applying for VISA. Thanks
in advance.
avatar
t*r
5
我H1还剩半个月,extension 现在还没批下来,家里去签B2会受影响吗?
avatar
m*m
6
rt
avatar
w*8
8
该来的总是要来的
avatar
l*0
9
chase的125, 弄DD晚了,没法赶在60天之前到账户,有什么补救办法?
谢谢,请问各位大侠指点。
avatar
s*a
10
我们的NOL中国大使馆显示3月份已发出但现在还没收到是怎么回事?
avatar
s*c
11
在国内至少呆半年以上再说吧

Thanks

【在 z*******o 的大作中提到】
: My parents have legally stayed in the U.S for about a year. Their I-94 was
: extented to 1 year. My parents want to come to U.S again right after going
: back to China. How long do they have to wait until applying for VISA. Thanks
: in advance.

avatar
E*V
12
Huan Jian SHu Meng

【在 m****m 的大作中提到】
: rt
avatar
g*g
13
我刚看了一下我跑的一个cluster, 随便拿个instance, 差不多3-5天就会health check
不响应,被shutdown,然后auto-scaling 再起一个。99.95%是不可能的,我看有99%差
不多,但你看小字就知道Amazon的定义不同。
不过单个instance的availability不是很重要。哪怕只有99%,一个cluster同时不
当超过一个的概率就有99.99%,所以不是关键。
就过去两年以我们公司的记录来看,AWS的region availability大约有99.95%,唯一一
次比较长的当是前年平安夜因为人为原因东部3个zone ELB大面积挂掉,导致整个
region挂掉。
如果你的目标是99.9%-99.95%,zone failover应该够了,就是一个region 3个zone平
均分配instance, 每个服务都至少3个zone里都有instance. 如果你的目标是99.95%-99
.99%,需要实现region failover. 东部挂了能自动换到西部。这个AWS的支持不好,要
自己实现一堆东西。我们的platform team花了一年的时间做这个。

【在 w**z 的大作中提到】
: 99.95% up?
: http://aws.amazon.com/ec2/sla/

avatar
p*y
14
要做最坏的打算了
avatar
z*u
15
用ING
ally paypal这些试试
死马当活马
avatar
s*u
16
取决于具体处理case 的VO.
可能1-2 days, 可能30-60 days
官僚体制就是这样

【在 s******a 的大作中提到】
: 我们的NOL中国大使馆显示3月份已发出但现在还没收到是怎么回事?
avatar
m*m
17
xiexie

【在 E*V 的大作中提到】
: Huan Jian SHu Meng
avatar
y*u
18
这个failover是在Zuul里做的吗?

check
99

【在 g*****g 的大作中提到】
: 我刚看了一下我跑的一个cluster, 随便拿个instance, 差不多3-5天就会health check
: 不响应,被shutdown,然后auto-scaling 再起一个。99.95%是不可能的,我看有99%差
: 不多,但你看小字就知道Amazon的定义不同。
: 不过单个instance的availability不是很重要。哪怕只有99%,一个cluster同时不
: 当超过一个的概率就有99.99%,所以不是关键。
: 就过去两年以我们公司的记录来看,AWS的region availability大约有99.95%,唯一一
: 次比较长的当是前年平安夜因为人为原因东部3个zone ELB大面积挂掉,导致整个
: region挂掉。
: 如果你的目标是99.9%-99.95%,zone failover应该够了,就是一个region 3个zone平
: 均分配instance, 每个服务都至少3个zone里都有instance. 如果你的目标是99.95%-99

avatar
s*n
19
还是那句话,砸就砸呗,就算到了2000点,最严重的后果是什么?
根本没什么啊,就是股市亏钱而已。金融体系好好的,实体经济基本不受影响。
avatar
w*u
20
没关系
我的就是80天左右才到了,还是给了我$125
avatar
d*f
22
那就是扯了,2000点真是要好多人跳楼了,股市的钱哪里来的,不都是非法集资么

【在 s******n 的大作中提到】
: 还是那句话,砸就砸呗,就算到了2000点,最严重的后果是什么?
: 根本没什么啊,就是股市亏钱而已。金融体系好好的,实体经济基本不受影响。

avatar
k*1
23
我的恐怕也要晚了。问题是只晚了3天而已。。。
我打算电话跟他们argue一下
avatar
w*z
24
我的另外一个projet 用了Eureka, 不错。
现在,要把我们公司的site (当初脑残用wordpress 搞的)放到EC2上,一个instance
应该不行的。但wordpress的image是存在disk上的。EBS volume 只能attach 在一个
instance. I found this blog,
http://blog.celingest.com/en/2013/09/10/highly-available-wordpr
Looks like I have to set some sort of distribute file system. like GlusterFS
mentioned in the blog.

【在 g*****g 的大作中提到】
: https://github.com/Netflix/eureka
avatar
s*n
25
跳楼是很小的问题吧,每天有一个人跳,跳一个月也许有点社会影响,
不过大多数普通老百姓还是看笑话而已。
再过一个月就开始谈论股市的成熟化,金融改革,长远是利大于弊。
不是有很多人说这是A股的1929重演嘛。

【在 d********f 的大作中提到】
: 那就是扯了,2000点真是要好多人跳楼了,股市的钱哪里来的,不都是非法集资么
avatar
l*e
26
应该5笔purchase也可以吧 赶紧刷卡买5个东西

【在 k*****1 的大作中提到】
: 我的恐怕也要晚了。问题是只晚了3天而已。。。
: 我打算电话跟他们argue一下

avatar
g*g
27
Typical solution for images is S3. S3's availability is quite high.
EBS also has lower availability based on experience. We are mostly using S3/
in-instance Cassandra cluster/RDS for storage.
Maybe all you need to do is changing the part of reading file to reading S3.

instance
GlusterFS

【在 w**z 的大作中提到】
: 我的另外一个projet 用了Eureka, 不错。
: 现在,要把我们公司的site (当初脑残用wordpress 搞的)放到EC2上,一个instance
: 应该不行的。但wordpress的image是存在disk上的。EBS volume 只能attach 在一个
: instance. I found this blog,
: http://blog.celingest.com/en/2013/09/10/highly-available-wordpr
: Looks like I have to set some sort of distribute file system. like GlusterFS
: mentioned in the blog.

avatar
D*9
28
2000点涨到5000点的钱谁赚取了?

【在 d********f 的大作中提到】
: 那就是扯了,2000点真是要好多人跳楼了,股市的钱哪里来的,不都是非法集资么
avatar
k*1
29
我去读了下条款
这个不行,非要DD

【在 l***e 的大作中提到】
: 应该5笔purchase也可以吧 赶紧刷卡买5个东西
avatar
w*z
30
You are right. WP has a plugin called total cache, so the images can be
stored on S3. But for the php files, need to write a script to sync whenever
they are changed (like installing new plugin etc.)
RDS is cool, can have multi-zone and read replicate. Don't need to worry
about that. The shared storage is pain.

S3/
S3.

【在 g*****g 的大作中提到】
: Typical solution for images is S3. S3's availability is quite high.
: EBS also has lower availability based on experience. We are mostly using S3/
: in-instance Cassandra cluster/RDS for storage.
: Maybe all you need to do is changing the part of reading file to reading S3.
:
: instance
: GlusterFS

avatar
w*n
31
那个是免月费的条款, 不是开户bonus

【在 l***e 的大作中提到】
: 应该5笔purchase也可以吧 赶紧刷卡买5个东西
avatar
g*g
32
Hmm, we don't deploy it that way. We would package the app into AMI and
deploy the same AMI to all instances. Php files can use in-instance storage.
On change, we create a new AMI, a new cluster, deploy the new AMI to the
new cluster and shut down the old one. We have it all set up with jenkins,
so it's as simple as running a jenkins job and in dev environment, a check-
in will trigger this jenkins job.
We never change files once deployed. If you change, the instance can be
killed and the new instance won't have it.

whenever

【在 w**z 的大作中提到】
: You are right. WP has a plugin called total cache, so the images can be
: stored on S3. But for the php files, need to write a script to sync whenever
: they are changed (like installing new plugin etc.)
: RDS is cool, can have multi-zone and read replicate. Don't need to worry
: about that. The shared storage is pain.
:
: S3/
: S3.

avatar
x*o
33
我的是开户100就好哎- -这还有区别么
avatar
w*z
34
We just started using AWS. For now, we only build AMI as base image (OS +
Apache + PHP etc.) In the case of Wordpress, we don't do active development
on that. We build a staging system and let developers to work on it, after
it's done, we flip it as production. If we manage to separate the media
storage from php file and want to use EC2 autoscaling, building AMI is a
good way to go.
For autoscaling, on top of the solution comes with AWS, you also have this
one:
http://techblog.netflix.com/2013/11/scryer-netflixs-predictive-
I guess you guys use both.

storage.

【在 g*****g 的大作中提到】
: Hmm, we don't deploy it that way. We would package the app into AMI and
: deploy the same AMI to all instances. Php files can use in-instance storage.
: On change, we create a new AMI, a new cluster, deploy the new AMI to the
: new cluster and shut down the old one. We have it all set up with jenkins,
: so it's as simple as running a jenkins job and in dev environment, a check-
: in will trigger this jenkins job.
: We never change files once deployed. If you change, the instance can be
: killed and the new instance won't have it.
:
: whenever

avatar
l*0
35
谢谢 各位的回复
我的125已经拿到手了 7天之后到账的
我弄的cashback 到的 没想到chase也算了DD 哈哈。。
希望给有类似情况的筒靴一个借鉴
avatar
g*g
36
We are using the Amazon auto scaling through Asgard. Scryer is something
that makes a difference when you have a 500 nodes cluster and you have years
of history and predictable pattern, with which you can save some cost. It's
probably overkill for most apps. A simple min/max/desired is enough for
capacity planning.

development

【在 w**z 的大作中提到】
: We just started using AWS. For now, we only build AMI as base image (OS +
: Apache + PHP etc.) In the case of Wordpress, we don't do active development
: on that. We build a staging system and let developers to work on it, after
: it's done, we flip it as production. If we manage to separate the media
: storage from php file and want to use EC2 autoscaling, building AMI is a
: good way to go.
: For autoscaling, on top of the solution comes with AWS, you also have this
: one:
: http://techblog.netflix.com/2013/11/scryer-netflixs-predictive-
: I guess you guys use both.

avatar
i*d
37
chase在给bonus这方面格外nice

【在 l*******0 的大作中提到】
: 谢谢 各位的回复
: 我的125已经拿到手了 7天之后到账的
: 我弄的cashback 到的 没想到chase也算了DD 哈哈。。
: 希望给有类似情况的筒靴一个借鉴

avatar
w*z
38
That makes sense.
Asgard, looks like a useful tool.

years
's

【在 g*****g 的大作中提到】
: We are using the Amazon auto scaling through Asgard. Scryer is something
: that makes a difference when you have a 500 nodes cluster and you have years
: of history and predictable pattern, with which you can save some cost. It's
: probably overkill for most apps. A simple min/max/desired is enough for
: capacity planning.
:
: development

相关阅读
logo
联系我们隐私协议©2024 redian.news
Redian新闻
Redian.news刊载任何文章,不代表同意其说法或描述,仅为提供更多信息,也不构成任何建议。文章信息的合法性及真实性由其作者负责,与Redian.news及其运营公司无关。欢迎投稿,如发现稿件侵权,或作者不愿在本网发表文章,请版权拥有者通知本网处理。