* [PATCH] test: fix vinyl/deferred_delete failure
@ 2019-03-15 15:57 Vladimir Davydov
2019-03-15 15:58 ` Vladimir Davydov
0 siblings, 1 reply; 2+ messages in thread
From: Vladimir Davydov @ 2019-03-15 15:57 UTC (permalink / raw)
To: tarantool-patches
The patch fixes the following test failure:
| --- vinyl/deferred_delete.result Fri Mar 15 18:17:05 2019
| +++ vinyl/deferred_delete.reject Fri Mar 15 18:18:18 2019
| @@ -577,7 +577,7 @@
| ...
| sk:stat().rows -- ditto
| ---
| -- 5
| +- 25
| ...
| s:drop()
| ---
The failure was introduced by commit 6dd0d2fb7e0e ("vinyl: do not apply
run_count_per_level to the last level"). Due to the commit compaction of
the secondary index may happen before compaction of the primary index,
in which case deferred DELETE statements won't make it to the secondary
index in time against the test expectation. Fix this by making the first
run big enough to prevent major compaction from kicking in.
Follow-up #3657
Closes #4047
---
test/vinyl/deferred_delete.result | 36 ++++++++++++++++++++++--------------
test/vinyl/deferred_delete.test.lua | 18 ++++++++++++------
2 files changed, 34 insertions(+), 20 deletions(-)
diff --git a/test/vinyl/deferred_delete.result b/test/vinyl/deferred_delete.result
index 61f81ce2..0a380ed5 100644
--- a/test/vinyl/deferred_delete.result
+++ b/test/vinyl/deferred_delete.result
@@ -503,6 +503,14 @@ pk = s:create_index('pk', {run_count_per_level = 10})
sk = s:create_index('sk', {run_count_per_level = 10, parts = {2, 'unsigned'}, unique = false})
---
...
+-- Write a run big enough to prevent major compaction from kicking in
+-- (run_count_per_level is ignored on the last level - see gh-3657).
+dummy_rows = 100
+---
+...
+for i = 1001, 1000 + dummy_rows do s:replace{i, i} end
+---
+...
for i = 1, 10 do s:replace{i, i} end
---
...
@@ -525,19 +533,19 @@ for i = 2, 10, 2 do s:replace{i, i * 100} end
box.commit()
---
...
-sk:select()
+sk:select({1000}, {iterator = 'le'})
---
-- - [2, 200]
- - [4, 400]
- - [6, 600]
+- - [10, 1000]
- [8, 800]
- - [10, 1000]
+ - [6, 600]
+ - [4, 400]
+ - [2, 200]
...
-pk:stat().rows -- 10 old REPLACEs + 5 DELETEs + 5 new REPLACEs
+pk:stat().rows - dummy_rows -- 10 old REPLACEs + 5 DELETEs + 5 new REPLACEs
---
- 20
...
-sk:stat().rows -- 10 old REPLACEs + 5 new REPLACEs
+sk:stat().rows - dummy_rows -- 10 old REPLACEs + 5 new REPLACEs
---
- 15
...
@@ -563,19 +571,19 @@ sk:compact()
while sk:stat().disk.compaction.count == 0 do fiber.sleep(0.001) end
---
...
-sk:select()
+sk:select({1000}, {iterator = 'le'})
---
-- - [2, 200]
- - [4, 400]
- - [6, 600]
+- - [10, 1000]
- [8, 800]
- - [10, 1000]
+ - [6, 600]
+ - [4, 400]
+ - [2, 200]
...
-pk:stat().rows -- 5 new REPLACEs
+pk:stat().rows - dummy_rows -- 5 new REPLACEs
---
- 5
...
-sk:stat().rows -- ditto
+sk:stat().rows - dummy_rows -- ditto
---
- 5
...
diff --git a/test/vinyl/deferred_delete.test.lua b/test/vinyl/deferred_delete.test.lua
index 93b5b358..240429fc 100644
--- a/test/vinyl/deferred_delete.test.lua
+++ b/test/vinyl/deferred_delete.test.lua
@@ -184,6 +184,12 @@ s:drop()
s = box.schema.space.create('test', {engine = 'vinyl'})
pk = s:create_index('pk', {run_count_per_level = 10})
sk = s:create_index('sk', {run_count_per_level = 10, parts = {2, 'unsigned'}, unique = false})
+
+-- Write a run big enough to prevent major compaction from kicking in
+-- (run_count_per_level is ignored on the last level - see gh-3657).
+dummy_rows = 100
+for i = 1001, 1000 + dummy_rows do s:replace{i, i} end
+
for i = 1, 10 do s:replace{i, i} end
box.snapshot()
@@ -193,10 +199,10 @@ for i = 1, 10, 2 do s:delete{i} end
for i = 2, 10, 2 do s:replace{i, i * 100} end
box.commit()
-sk:select()
+sk:select({1000}, {iterator = 'le'})
-pk:stat().rows -- 10 old REPLACEs + 5 DELETEs + 5 new REPLACEs
-sk:stat().rows -- 10 old REPLACEs + 5 new REPLACEs
+pk:stat().rows - dummy_rows -- 10 old REPLACEs + 5 DELETEs + 5 new REPLACEs
+sk:stat().rows - dummy_rows -- 10 old REPLACEs + 5 new REPLACEs
-- Compact the primary index to generate deferred DELETEs.
box.snapshot()
@@ -208,10 +214,10 @@ box.snapshot()
sk:compact()
while sk:stat().disk.compaction.count == 0 do fiber.sleep(0.001) end
-sk:select()
+sk:select({1000}, {iterator = 'le'})
-pk:stat().rows -- 5 new REPLACEs
-sk:stat().rows -- ditto
+pk:stat().rows - dummy_rows -- 5 new REPLACEs
+sk:stat().rows - dummy_rows -- ditto
s:drop()
--
2.11.0
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] test: fix vinyl/deferred_delete failure
2019-03-15 15:57 [PATCH] test: fix vinyl/deferred_delete failure Vladimir Davydov
@ 2019-03-15 15:58 ` Vladimir Davydov
0 siblings, 0 replies; 2+ messages in thread
From: Vladimir Davydov @ 2019-03-15 15:58 UTC (permalink / raw)
To: tarantool-patches
On Fri, Mar 15, 2019 at 06:57:51PM +0300, Vladimir Davydov wrote:
> The patch fixes the following test failure:
>
> | --- vinyl/deferred_delete.result Fri Mar 15 18:17:05 2019
> | +++ vinyl/deferred_delete.reject Fri Mar 15 18:18:18 2019
> | @@ -577,7 +577,7 @@
> | ...
> | sk:stat().rows -- ditto
> | ---
> | -- 5
> | +- 25
> | ...
> | s:drop()
> | ---
>
> The failure was introduced by commit 6dd0d2fb7e0e ("vinyl: do not apply
> run_count_per_level to the last level"). Due to the commit compaction of
> the secondary index may happen before compaction of the primary index,
> in which case deferred DELETE statements won't make it to the secondary
> index in time against the test expectation. Fix this by making the first
> run big enough to prevent major compaction from kicking in.
>
> Follow-up #3657
> Closes #4047
> ---
> test/vinyl/deferred_delete.result | 36 ++++++++++++++++++++++--------------
> test/vinyl/deferred_delete.test.lua | 18 ++++++++++++------
> 2 files changed, 34 insertions(+), 20 deletions(-)
Trivial. Pushed to 2.1.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2019-03-15 15:58 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-03-15 15:57 [PATCH] test: fix vinyl/deferred_delete failure Vladimir Davydov
2019-03-15 15:58 ` Vladimir Davydov
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox