BUG: groupby.sum removed columns in case sequential calls of several groupby.sum

[prutskov]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the master branch of pandas.

Reproducible Example

import pandas as pd
from pandas import util

df = util.testing.makeMixedDataFrame()
df = df.to_numpy()
df = pd.DataFrame(df)
df.columns = ["A", "B", "C", "D"]
df = df.set_index("B")

df1, df2 = df.iloc[:2], df.iloc[2:]

groupby1 = df1.groupby("B").sum()
groupby2 = df2.groupby("B").sum()

df_conc = pd.concat([groupby1, groupby2], axis=0)
print(f'\nConcatenated dataframe from partial results:\n{df_conc}')
print(f'\nOriginal dataframe:\n{df}')

conc_sum = df_conc.groupby(level=0).sum()
orig_sum = df.groupby(level=0).sum()

print(f'\nConcatenated df groupby.sum: \n{conc_sum}')
print(f'\nOriginal df groupby.sum: \n{orig_sum}')

Issue Description

Script output

Concatenated dataframe from partial results:
       A         C
B
0.0  0.0      foo1
1.0  1.0      foo2
0.0  6.0  foo3foo5
1.0  3.0      foo4

Original dataframe:
       A     C          D
B
0.0  0.0  foo1 2009-01-01
1.0  1.0  foo2 2009-01-02
0.0  2.0  foo3 2009-01-05
1.0  3.0  foo4 2009-01-06
0.0  4.0  foo5 2009-01-07

Concatenated df groupby.sum:
       A    # Column `C` is missed
B
0.0  6.0
1.0  4.0

Original df groupby.sum:
       A             C
B
0.0  6.0  foo1foo3foo5
1.0  4.0      foo2foo4

groupby.sum misses columns in case of splitting DataFrame on several parts and applying groupby.sum first for each part then for concatenation result of processed parts.

One observation: groupby.sum changed dtypes of partitions. If align dtypes of partitions with df.dtypes we can get expected behavior.

Expected Behavior

Expected results:

Concatenated df groupby.sum:
       A             C
B
0.0  6.0  foo1foo3foo5
1.0  4.0      foo2foo4

Original df groupby.sum:
       A             C
B
0.0  6.0  foo1foo3foo5
1.0  4.0      foo2foo4

Installed Versions

INSTALLED VERSIONS

commit : 73c6825
python : 3.8.10.final.0
python-bits : 64
OS : Linux
OS-release : 5.4.0-65-generic
Version : #73-Ubuntu SMP Mon Jan 18 17:25:17 UTC 2021
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US
LOCALE : en_US.ISO8859-1

pandas : 1.3.3
numpy : 1.21.0
pytz : 2021.1
dateutil : 2.8.1
pip : 21.1.3
setuptools : 52.0.0.post20210125
Cython : None
pytest : 6.2.4
hypothesis : None
sphinx : 4.1.2
blosc : None
feather : 0.4.1
xlsxwriter : None
lxml.etree : 4.6.3
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.0.1
IPython : 7.25.0
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : 2021.06.1
fastparquet : None
gcsfs : None
matplotlib : 3.2.2
numexpr : 2.7.3
odfpy : None
openpyxl : 3.0.7
pandas_gbq : 0.15.0
pyarrow : 4.0.1
pyxlsb : None
s3fs : 2021.06.1
scipy : 1.7.0
sqlalchemy : 1.4.20
tables : 3.6.1
tabulate : None
xarray : 0.18.2
xlrd : 2.0.1
xlwt : None
numba : None

[jbrockmendel]

This works on main. Maybe could use a test (or closeable)

[rhshadrach]

This is tested as part of #46560