1. 基本測試
2. 功能測試: JDBC -> Stream -> HDFS
3. 安裝測試:Install case
@oharastream/developer
If you have found any issue or problems about version 0.9, please add the issue number here.
@oharastream/frontend 我在docker(ssh) mode裡面加入節點後沒看到資源顯示,但我透過APIs去敲是可以看到的


~加入節點後,回到建立workspace的頁面卻沒有任何節點顯示~


更正:
是我誤會的UI操作邏輯 :( 沒事沒事
加入plugins後沒有顯示內容(例如有哪些connectors, streams)

connector的線已經連結到一個topic,但在connector的設定裡面卻依然沒有顯示已經選擇的topic


connector的線已經連結到一個topic,但在connector的設定裡面卻依然沒有顯示已經選擇的topic
It seems duplicated to #3693? /cc @wu87988622
duration型態的預設值不見了

加入plugins後沒有顯示內容(例如有哪些connectors, streams)
fail to observe configurator log from UI (but API works well)

乍看之下很像是UI嘗試從"已經加入的節點"找configurator log(?),可是configurator可能根本不在已經加入的節點列表裡面。或許UI應該直接打${configurator_url}/logs/configurator就好,而不用去在意原本的nodes列表
fail to observe configurator log from UI (but API works well)
乍看之下很像是UI嘗試從"已經加入的節點"找configurator log(?),可是configurator可能根本不在已經加入的節點列表裡面。或許UI應該直接打${configurator_url}/logs/configurator就好,而不用去在意原本的nodes列表
I tested the same behavior in our Demo env with 0.9, and it works well:

@saivirtue 你嘗試看看讓configurator不是位於nodes列表 (如我上面的敘述)
例如node00跑configurator,然後只把node01加入configurator,並把其他服務啟動在node01
@saivirtue 你嘗試看看讓configurator不是位於nodes列表 (如我上面的敘述)
例如node00跑configurator,然後只把node01加入configurator,並把其他服務啟動在node01
It seems like something wrong in docker mode
I will update the test result later...
@saivirtue 你嘗試看看讓configurator不是位於nodes列表 (如我上面的敘述)
例如node00跑configurator,然後只把node01加入configurator,並把其他服務啟動在node01
The reason UI displays "nothing" is because of API return "empty log".
The default sinceSeconds in UI is using 600, but your native API is using empty (which means all data)
BTW, we use the node name from logs.hostname of API response, node list doesn't participate in it :)
Configurator log should be able to fetch even if no workspace

The default sinceSeconds in UI is using 600, but your native API is using empty (which means all data)
看起來應該是APIs的錯誤 (see https://github.com/oharastream/ohara/issues/3770)
Configurator log should be able to fetch even if no workspace
這個我剛剛也有遇到,一定要先建立才可以看logs :(
topic (pipeline only)的設定按鈕按了沒有反應 (當時沒有任何運作中的connectors, streams)

pipeline上的connector已經啟動了(connector and pipeline APIs都顯示running),但是畫面依然顯示stopped
ps. 已經重複刷新頁面,依然是stopped



首頁的文字typo:
Ohara is a scalable streaming platform that allows users to easily organized their input, output, and streaming applications with a clean and comprehensive GUI
"into the worksapce" 無法點擊

duration型態的資料存放有誤,應該是"1 seconds"而不是"1"


首頁的文字typo:
Not that good at grammar, but I believe verbs after the "to" should be "Organize" not "Organized" :)
"into the worksapce" 無法點擊
It's intended. Because it's the current workspace
topic (pipeline only)的設定按鈕按了沒有反應 (當時沒有任何運作中的connectors, streams)
Hmm, the config button should be removed from topics
If there is one workspace only, the "into the workspace" means current workspace (which should be disabled here).
Maybe we can provide a tooltip as a hint like: "This is the workspace that you're at now"?
If there is one workspace only, the "into the workspace" means current workspace (which should be disabled here).
Maybe we can provide a tooltip as a hint like: "This is the workspace that you're at now"?
+1
Maybe we can provide a tooltip as a hint like: "This is the workspace that you're at now"?
/cc @eechih @wu87988622 for feedback on this?
node 編輯頁面不應該可以修改hostname (會產生新的node)

Maybe we can provide a tooltip as a hint like: "This is the workspace that you're at now"?
+1
我在docker(ssh) mode裡面加入節點後沒看到資源顯示,但我透過APIs去敲是可以看到的
這是APIs的問題 (#3785)



從pipeline-only topic無法拉線連接到JsonOut
Updated: Cannot be reproduced

Type.Duration的預設值沒有顯示 (#4098)
Quickstart VM --> Nodes --> ADD NODE --> 填入正確資料 --> ADD

Update:
- Memory 8G --> 12G 這個情形就消失了.
QuickstartVM(Mem 12GB), 操作步驟:
issue #4099
Display name is not capitalized
*first

Issue: #4151
We should confirm before deleting

Thanks for reporting @eechih! The issue is tracked in #4152
Error "Failed to inspect topic XX Info" when open new window(#4109)

重開所有服務後,event logs裡面依舊有前世的記憶

錯誤訊息不正確
Update: required by #4124

不存在的key但允許儲存,儲存完後再開會變成另外一個key

重現方式:
Update: required by #4124

pipelinename應該換成拼法正確的字,或者加上編號的單字 (如pipeline0)
Thanks for raising this. The issue is tracked in #4153
點選"delete this pipeline"後並沒有幫忙刪除該pipeline上原有的元件
重現方式:
Issue is tracked in #4140
在 connector 因後端出現錯誤而進入 FAILED 狀態下, 若使用者需要修改 property, 目前操作流程需要先將 connector stop, 才能 edit property.
因此, 在 FAILED 時應該要 disable "Edit" button.
Issue is tracked in #4173
Log 內容自動跳到最前面
操作順序: Develop Tools -> LOGS -> select worker
fixed in #4262
有時候會出現下列情形(目前尚無法重現), 目前遇到 4-5 次:
在 Develop Tools, 從 LOGS tab 切換至 TOPICS tab 後即跳至 500 page:
500: Oops, something went terribly wrong!
You either tried some shady route or you came here by mistake. Whichever it is, try using the navigation
refresh page 後似乎沒遇到什麼問題~
Update: 應該抓到了, 請見 #4121
有時候會出現下列情形(目前尚無法重現), 目前遇到 4-5 次:
在 Develop Tools, 從 LOGS tab 切換至 TOPICS tab 後即跳至 500 page:500: Oops, something went terribly wrong!
You either tried some shady route or you came here by mistake. Whichever it is, try using the navigationrefresh page 後似乎沒遇到什麼問題~
This may be related to #4091, could you pull the PR and test the same scenario again?

The error.folder setting in FtpSource is defined with Optional, but the backend require it...
Failed to inspect topic XX info after delete topic(#4139)
FTP Sink 的 File Need Header 無法打勾

[Updated by sam]
ping @eechih
This issue should be fixed in #4187, could you help me to verify after merged?
重開所有服務後,event logs裡面依舊有前世的記憶
@chia7712 我會在下個版本處理 #4143
Minor: a vertical line in the TOPICS tab.

從pipeline-only topic無法拉線連接到JsonOut
@chia7712 Are you still experiencing this? I'm unable to repro this on my machine...
Are you still experiencing this? I'm unable to repro this on my machine...
I fail to produce this error :(
Failed to create shared topic
Describe the bug
建立 Shared Topic 遇到啟動失敗後,沒有出現在 Toolbox 或 Topics,也無法再次建立(使用相同名稱)。
To Reproduce



related to #4175
沒有加入到 Pipeline 的 Topic,不應該出現在選單中

Issue is tracked in #4176
調整 number of task 後,存檔會失敗


related to #4177
I fail to produce this error :(
Not problem. I will mark the issue as resolved then :)
啟動中的 Stream,如果 Save Change 會造成 Stream 的狀態變成 Failed

related to #4178
Minor: Event log 訊息太長時, row 高度會暴增, 建議將訊息截斷


port的顯示不用逗號,那個不是薪水

已經啟動的connector依然可以再次啟動?
This issue is tracked in #4236

sink connector的nodes和metrics沒有顯示出來
Fixed in #4196
sink connector的nodes和metrics沒有顯示出來
@chia7712 A fix is on the way... #4196
cc @wu87988622 Please mark the comment as resolve once the PR is in. Thanks :)
已經啟動的connector依然可以再次啟動?
@eechih I think we have missed this... So I will file another issue to fix this 😅
已經啟動的connector依然可以再次啟動?
@eechih I think we have missed this... So I will file another issue to fix this 😅
Ok~ Thanks @konekoya

issues #4247
Pipelines 清單自動只剩一個. refresh page 一開始有三個 pipeline:

靜置約 10 秒剩一個:

重現: 建一個 workspace, 接著再建三個 pipelines, 過一會兒應該就可發現
fix in #4148
重現: 建一個 workspace, 接著再建三個 pipelines, 過一會兒應該就可發現
@wu87988622 it's happing again... 🙇
FtpSourceConnector 相關.
重現:
error: java.util.NoSuchElementException: /vito/source/files doesn't exist
at oharastream.ohara.client.filesystem.ftp.FtpClient$Builder$FtpClientImpl.fileType(FtpClient.scala:497)
at oharastream.ohara.client.filesystem.ftp.FtpClient$Builder$$anon$1.$anonfun$fileType$1(FtpClient.scala:304)
at oharastream.ohara.client.filesystem.ftp.FtpClient$Builder$$anon$1.retry(FtpClient.scala:281)
at oharastream.ohara.client.filesystem.ftp.FtpClient$Builder$$anon$1.fileType(FtpClient.scala:304)
at oharastream.ohara.client.filesystem.ftp.FtpFileSystem$Builder$FtpFileSystemImpl.$anonfun$fileType$1(FtpFileSystem.scala:281)
at oharastream.ohara.client.filesystem.FileSystem.wrap(FileSystem.scala:92)
at oharastream.ohara.client.filesystem.FileSystem.wrap$(FileSystem.scala:90)
at oharastream.ohara.client.filesystem.ftp.FtpFileSystem$Builder$FtpFileSystemImpl.wrap(FtpFileSystem.scala:136)
at
: : : :
==> 資料夾不存在但拋 NoSuchElementException 似乎有點奇怪?
metrics 的數字與單位, 距離似乎太遙遠?

Tracked in #4255
在 Property view -> Metrics 打開情形下, 若關閉 Property view 後再度打開, metrics 會處於關閉狀態.
希望能記住之前是否打開, 下次 Property view 打開時可以恢復之前的狀態
Tracked in #4256
metrics 目前是反應後端 runtime 下的數字, 因此 pipeline component 在 stop 的狀態下, metrics 是否要歸 0 比較合理?
Internal Topic 的錯誤, 除了 name 之外, 可能還是附上 pipeline 上的名稱, 會比較清楚~
Ex: Failed to stop topic 59sn5fcc4y(T1)

Tracked in #4254
metrics 目前是反應後端 runtime 下的數字, 因此 pipeline component 在 stop 的狀態下, metrics 是否要歸 0 比較合理?
@vitojeng 目前會做清除,但是要等到下一次更新,大約是等待五秒後,就會清掉了
Internal Topic 的錯誤, 除了 name 之外, 可能還是附上 pipeline 上的名稱, 會比較清楚~
Ex: Failed to stop topic 59sn5fcc4y(T1)
@vitojeng Do you mean it should display __T1__ instead of __59sn5fcc4y__? The latter one actually doesn't mean anything to our users, so we always hide it from the UI :)
metrics 的數字與單位, 距離似乎太遙遠?
I think it's okay if we put them together anyway. Like 5,998 messages instead of breaking them apart
Pinging @eechih for idea :)
在 Property view -> Metrics 打開情形下, 若關閉 Property view 後再度打開, metrics 會處於關閉狀態.
希望能記住之前是否打開, 下次 Property view 打開時可以恢復之前的狀態
Pinging @eechih and everyone for thoughts @oharastream/frontend
資料夾不存在但拋 NoSuchElementException 似乎有點奇怪?
Or IllegalArgumentException?
metrics 的數字與單位, 距離似乎太遙遠?
Internal Topic 的錯誤, 除了 name 之外, 可能還是附上 pipeline 上的名稱, 會比較清楚~
Ex: Failed to stop topic 59sn5fcc4y(T1)
@vitojeng @konekoya 這兩個問題不影響使用者操作,等到下個版本再來討論,OK?
@vitojeng @konekoya 這兩個問題不影響使用者操作,等到下個版本再來討論,OK?
OK~ 有需要 backport 的我會再額外提醒
資料夾不存在但拋 NoSuchElementException 似乎有點奇怪?
Or IllegalArgumentException?
比 NoSuchElementException 好 10 倍

死掉的節點不應該讓使用者能選擇
fix in #4251
@vitojeng @konekoya 這兩個問題不影響使用者操作,等到下個版本再來討論,OK?
No problem :)
Env:
Step:
在 pipeline1 切換到 pipeline 2 時, 出現 error: Failed to stop topic XXXXX
而pipeline2 切到 pipeline 1 時, 沒有出現 error.
不知為何在切換 pipeline 時會需要做這個動作?

Detail

確認後若不是太困難, 希望這個 issue 可以 backport 到 0.9.x :)
確認後若不是太困難, 希望這個 issue 可以 backport 到 0.9.x :)
Vito, there's already a PR addresses this issue. I'll try to get it in today and ping you again for testing. (Sorry, it seems like GitHub's mention feature is broken so I somehow couldn't tag you in this comment)
related with https://github.com/oharastream/ohara/issues/3706#issuecomment-595061071
dead node 的資訊顯示怪怪的~


fix #4263
觀察 Snackbar 目前容易被忽略的因素: 一個是顯示在左下角, 另一個是顯示時間不長.
目前 layout 大部份操作都集中在畫面上半部, 使用者目光停留在上半部機會很高. 在高解晰度的大螢幕下, 左下角的 Snackbar 相信被忽略的機會大增. 因此建議 Snackbar 顯示在畫面中上方, 應該可以增加被使用者注意到的機會~ (希望可以 backport 到 0.9.x)
Fixed in #4268
在 Property view 開啟的情形下, 從 Developer Tools/LOGS 切換到 Developer Tools/Topic 時, Property view 會自動被關閉.
確認後若不是太困難, 希望這個 issue 可以 backport 到 0.9.x :)
@vitojeng The PR has been merged. But It still needs some time to build the image :)
觀察 Snackbar 目前容易被忽略的因素: 一個是顯示在左下角, 另一個是顯示時間不長.
目前 layout 大部份操作都集中在畫面上半部, 使用者目光停留在上半部機會很高. 在高解晰度的大螢幕下, 左下角的 Snackbar 相信被忽略的機會大增. 因此建議 Snackbar 顯示在畫面中上方, 應該可以增加被使用者注意到的機會~ (希望可以 backport 到 0.9.x)
Pinging @oharastream/frontend for thoughts
related with #3706 (comment)
dead node 的資訊顯示怪怪的~
@vitojeng can you get me unavailable node response, i can't reproduce that, thanks
fix in #4263
@vitojeng can you get me unavailable node response, i can't reproduce that, thanks
Just shutdown the one of kubernetes slave node and you should be able to reproduce it.

toolbox被log畫面擋住,沒有卷軸可以使用
Issue tracked in #4267
log畫面開著的時候,pipeline上的"變更"都會觸發重拉log...這是預期的行為嗎?
log畫面開著的時候,pipeline上的"變更"都會觸發重拉log...這是預期的行為嗎?
DevTool工具打開的時候,Pipeline上選中的component會觸發搜尋對應的log / topicData:
DevTool - topic tab :
DevTool - log tab :
@saivirtue 了解,就是之前討論的選種哪個物件就找尋可能的logs
@saivirtue 了解,就是之前討論的選種哪個物件就找尋可能的logs
You are right :)

如果現在UI嚴格限制只能1-to-1的話,那在連線後就應該把"連線"的按鈕遮掉...
Will be fixed in #4278
同一個workspace,不同pipeline裡面新增一個重複名稱的物件,會顯示錯誤但該物件依然留在畫布上
@chia7712 Hmm, I'm not able to repro this. Not sure if I understand this right:
This works okay for me
@konekoya it seems a temporary connection error so I remove this comment :)
在 Create Pipeline 時, 若使用重覆的 name, 會在左下角顯示 "Failed to create pipeline XXX".
比較好的做法應該是 Create Pipeline Dialog 時通知使用者修正. (0.10 only)
Tracked in #4286
在 Edit Property Dialog, 我們使用 * 顯示 password; 但在 Property View 裡並沒有使用 *:

Please backport to 0.9.x
目前 Metrics 開關, 似乎是所有 Pipeline 都共用同一組.
每個 Pipeline , 應該有自己的 Metrics 開關, 預設是關. (0.10 only)
討論: Pipeline componet name 目前只能小寫與數字.
使用者在想要以多個單字組合來命名時, 全部小寫會難以辦認, 可能需要討論如何修正.(0.10 only)

(我好像之前回報過) unavailable nodes should NOT be selectable!
Tracked in #4313
討論: Pipeline componet name 目前只能小寫與數字.
因為k8s對名稱有強烈要求,為了一致性所以讓所有物件都套用相同限制

確認一下,錯誤訊息跑到右上角是預期的行為嗎?
確認一下,錯誤訊息跑到右上角是預期的行為嗎?
Yep. This was recently changed in #4268 :)

欄位"c"跑版了
Tracked in #4306
欄位"c"跑版了
IIRC, the "latest column" will align field to right.
/cc @konekoya Am I right?

~應該要禁止加入重複的節點~
更正,該節點並沒有成功被加入到後端,不過畫面上卻有顯示...
補充,嘗試刪除"多出來的節點":

tracked in #4303
觀察 Response, API(/connectors) 裡 A 的 taskStatus 是空的, C 的 taskStatus 有值
connector.state有值嗎? 這可能是開connector成功,但開task的時候失敗。可以看一下worker logs
IIRC, the "latest column" will align field to right.
/cc @konekoya Am I right?
Right, but the cell data should also align to the right. I can submit a PR to fix this if you're busy with the API refactoring :)
Node unavailable 時:
- Fixed in #4313Alive(綠色) or Dead(紅色) #4368
(我好像之前回報過) unavailable nodes should NOT be selectable!
@chia7712 Thanks for reporting, I will look into it today (for both node related issues) :)
CPU/Memory: NaN 應該顯示 -
@vitojeng This will be fixed in #4313 :)
前幾天有遇到, 現在又遇到...但重現不出來 QQ

updated: Due to multiple tasks of connector's metrics, not a bug
更正,該節點並沒有成功被加入到後端,不過畫面上卻有顯示...
補充,嘗試刪除"多出來的節點":
刪除本尊 -> 成功
刪除複製人 -> 失敗,因為刪除的按鈕不能按
@chia7712 I'm unable to repro this. Adding two same name nodes is blocked by the APIs
I'm unable to repro this. Adding two same name nodes is blocked by the APIs
I will report this issue if I find a way to reproduce this bug :)
Node unavailable 時:
CPU/Memory: NaN 應該顯示 -
State: 應與其他地方一致, 顯示 Alive(綠色) or Dead(紅色)
@oharastream/frontend Is there a issue ?
CPU/Memory: NaN 應該顯示
Fixed in #4313 :)
State: 應與其他地方一致, 顯示 Alive(綠色) or Dead(紅色)
Tracked in #4368
all issues are resolved so close this issue now.
feel free to raise issue on #3019
FYI @oharastream/developer
Most helpful comment
all issues are resolved so close this issue now.
feel free to raise issue on #3019
FYI @oharastream/developer